| @@ -1,1532 +1,1564 @@ | ||
| 1 | -<?php | |
| 2 | - | |
| 3 | -use BadMethodCallException; | |
| 4 | -use InvalidArgumentException; | |
| 5 | - | |
| 6 | -/** | |
| 7 | - * This class provides functions for converting CSS styles into inline style attributes in your HTML code. | |
| 8 | - * | |
| 9 | - * For more information, please see the README.md file. | |
| 10 | - * | |
| 11 | - * @version 1.2.0 | |
| 12 | - * | |
| 13 | - * @author Cameron Brooks | |
| 14 | - * @author Jaime Prado | |
| 15 | - * @author Oliver Klee <typo3-coding@oliverklee.de> | |
| 16 | - * @author Roman Ožana <ozana@omdesign.cz> | |
| 17 | - * @author Sander Kruger <s.kruger@invessel.com> | |
| 18 | - */ | |
| 19 | -class Emogrifier { | |
| 20 | - /** | |
| 21 | - * @var int | |
| 22 | - */ | |
| 23 | - const CACHE_KEY_CSS = 0; | |
| 24 | - | |
| 25 | - /** | |
| 26 | - * @var int | |
| 27 | - */ | |
| 28 | - const CACHE_KEY_SELECTOR = 1; | |
| 29 | - | |
| 30 | - /** | |
| 31 | - * @var int | |
| 32 | - */ | |
| 33 | - const CACHE_KEY_XPATH = 2; | |
| 34 | - | |
| 35 | - /** | |
| 36 | - * @var int | |
| 37 | - */ | |
| 38 | - const CACHE_KEY_CSS_DECLARATIONS_BLOCK = 3; | |
| 39 | - | |
| 40 | - /** | |
| 41 | - * @var int | |
| 42 | - */ | |
| 43 | - const CACHE_KEY_COMBINED_STYLES = 4; | |
| 44 | - | |
| 45 | - /** | |
| 46 | - * for calculating nth-of-type and nth-child selectors | |
| 47 | - * | |
| 48 | - * @var int | |
| 49 | - */ | |
| 50 | - const INDEX = 0; | |
| 51 | - | |
| 52 | - /** | |
| 53 | - * for calculating nth-of-type and nth-child selectors | |
| 54 | - * | |
| 55 | - * @var int | |
| 56 | - */ | |
| 57 | - const MULTIPLIER = 1; | |
| 58 | - | |
| 59 | - /** | |
| 60 | - * @var string | |
| 61 | - */ | |
| 62 | - const ID_ATTRIBUTE_MATCHER = '/(\\w+)?\\#([\\w\\-]+)/'; | |
| 63 | - | |
| 64 | - /** | |
| 65 | - * @var string | |
| 66 | - */ | |
| 67 | - const CLASS_ATTRIBUTE_MATCHER = '/(\\w+|[\\*\\]])?((\\.[\\w\\-]+)+)/'; | |
| 68 | - | |
| 69 | - /** | |
| 70 | - * @var string | |
| 71 | - */ | |
| 72 | - const CONTENT_TYPE_META_TAG = '<meta http-equiv="Content-Type" content="text/html; charset=utf-8">'; | |
| 73 | - | |
| 74 | - /** | |
| 75 | - * @var string | |
| 76 | - */ | |
| 77 | - const DEFAULT_DOCUMENT_TYPE = '<!DOCTYPE html>'; | |
| 78 | - | |
| 79 | - /** | |
| 80 | - * @var string | |
| 81 | - */ | |
| 82 | - private $html = ''; | |
| 83 | - | |
| 84 | - /** | |
| 85 | - * @var string | |
| 86 | - */ | |
| 87 | - private $css = ''; | |
| 88 | - | |
| 89 | - /** | |
| 90 | - * @var bool[] | |
| 91 | - */ | |
| 92 | - private $excludedSelectors = []; | |
| 93 | - | |
| 94 | - /** | |
| 95 | - * @var string[] | |
| 96 | - */ | |
| 97 | - private $unprocessableHtmlTags = ['wbr']; | |
| 98 | - | |
| 99 | - /** | |
| 100 | - * @var bool[] | |
| 101 | - */ | |
| 102 | - private $allowedMediaTypes = ['all' => true, 'screen' => true, 'print' => true]; | |
| 103 | - | |
| 104 | - /** | |
| 105 | - * @var mixed[] | |
| 106 | - */ | |
| 107 | - private $caches = [ | |
| 108 | - self::CACHE_KEY_CSS => [], | |
| 109 | - self::CACHE_KEY_SELECTOR => [], | |
| 110 | - self::CACHE_KEY_XPATH => [], | |
| 111 | - self::CACHE_KEY_CSS_DECLARATIONS_BLOCK => [], | |
| 112 | - self::CACHE_KEY_COMBINED_STYLES => [], | |
| 113 | - ]; | |
| 114 | - | |
| 115 | - /** | |
| 116 | - * the visited nodes with the XPath paths as array keys | |
| 117 | - * | |
| 118 | - * @var \DOMElement[] | |
| 119 | - */ | |
| 120 | - private $visitedNodes = []; | |
| 121 | - | |
| 122 | - /** | |
| 123 | - * the styles to apply to the nodes with the XPath paths as array keys for the outer array | |
| 124 | - * and the attribute names/values as key/value pairs for the inner array | |
| 125 | - * | |
| 126 | - * @var string[][] | |
| 127 | - */ | |
| 128 | - private $styleAttributesForNodes = []; | |
| 129 | - | |
| 130 | - /** | |
| 131 | - * Determines whether the "style" attributes of tags in the the HTML passed to this class should be preserved. | |
| 132 | - * If set to false, the value of the style attributes will be discarded. | |
| 133 | - * | |
| 134 | - * @var bool | |
| 135 | - */ | |
| 136 | - private $isInlineStyleAttributesParsingEnabled = true; | |
| 137 | - | |
| 138 | - /** | |
| 139 | - * Determines whether the <style> blocks in the HTML passed to this class should be parsed. | |
| 140 | - * | |
| 141 | - * If set to true, the <style> blocks will be removed from the HTML and their contents will be applied to the HTML | |
| 142 | - * via inline styles. | |
| 143 | - * | |
| 144 | - * If set to false, the <style> blocks will be left as they are in the HTML. | |
| 145 | - * | |
| 146 | - * @var bool | |
| 147 | - */ | |
| 148 | - private $isStyleBlocksParsingEnabled = true; | |
| 149 | - | |
| 150 | - /** | |
| 151 | - * Determines whether elements with the `display: none` property are | |
| 152 | - * removed from the DOM. | |
| 153 | - * | |
| 154 | - * @var bool | |
| 155 | - */ | |
| 156 | - private $shouldKeepInvisibleNodes = true; | |
| 157 | - | |
| 158 | - /** | |
| 159 | - * @var string[] | |
| 160 | - */ | |
| 161 | - private $xPathRules = [ | |
| 162 | - // child | |
| 163 | - '/\\s*>\\s*/' => '/', | |
| 164 | - // adjacent sibling | |
| 165 | - '/\\s+\\+\\s+/' => '/following-sibling::*[1]/self::', | |
| 166 | - // descendant | |
| 167 | - '/\\s+(?=.*[^\\]]{1}$)/' => '//', | |
| 168 | - // :first-child | |
| 169 | - '/([^\\/]+):first-child/i' => '*[1]/self::\\1', | |
| 170 | - // :last-child | |
| 171 | - '/([^\\/]+):last-child/i' => '*[last()]/self::\\1', | |
| 172 | - // attribute only | |
| 173 | - '/^\\[(\\w+|\\w+\\=[\'"]?\\w+[\'"]?)\\]/' => '*[@\\1]', | |
| 174 | - // attribute | |
| 175 | - '/(\\w)\\[(\\w+)\\]/' => '\\1[@\\2]', | |
| 176 | - // exact attribute | |
| 177 | - '/(\\w)\\[(\\w+)\\=[\'"]?([\\w\\s]+)[\'"]?\\]/' => '\\1[@\\2="\\3"]', | |
| 178 | - // element attribute~= | |
| 179 | - '/([\\w\\*]+)\\[(\\w+)[\\s]*\\~\\=[\\s]*[\'"]?([\\w-_\\/]+)[\'"]?\\]/' => '\\1[contains(concat(" ", @\\2, " "), concat(" ", "\\3", " "))]', | |
| 180 | - // element attribute^= | |
| 181 | - '/([\\w\\*]+)\\[(\\w+)[\\s]*\\^\\=[\\s]*[\'"]?([\\w-_\\/]+)[\'"]?\\]/' => '\\1[starts-with(@\\2, "\\3")]', | |
| 182 | - // element attribute*= | |
| 183 | - '/([\\w\\*]+)\\[(\\w+)[\\s]*\\*\\=[\\s]*[\'"]?([\\w-_\\s\\/:;]+)[\'"]?\\]/' => '\\1[contains(@\\2, "\\3")]', | |
| 184 | - // element attribute$= | |
| 185 | - '/([\\w\\*]+)\\[(\\w+)[\\s]*\\$\\=[\\s]*[\'"]?([\\w-_\\s\\/]+)[\'"]?\\]/' => '\\1[substring(@\\2, string-length(@\\2) - string-length("\\3") + 1) = "\\3"]', | |
| 186 | - // element attribute|= | |
| 187 | - '/([\\w\\*]+)\\[(\\w+)[\\s]*\\|\\=[\\s]*[\'"]?([\\w-_\\s\\/]+)[\'"]?\\]/' => '\\1[@\\2="\\3" or starts-with(@\\2, concat("\\3", "-"))]', | |
| 188 | - ]; | |
| 189 | - | |
| 190 | - /** | |
| 191 | - * Determines whether CSS styles that have an equivalent HTML attribute | |
| 192 | - * should be mapped and attached to those elements. | |
| 193 | - * | |
| 194 | - * @var bool | |
| 195 | - */ | |
| 196 | - private $shouldMapCssToHtml = false; | |
| 197 | - | |
| 198 | - /** | |
| 199 | - * This multi-level array contains simple mappings of CSS properties to | |
| 200 | - * HTML attributes. If a mapping only applies to certain HTML nodes or | |
| 201 | - * only for certain values, the mapping is an object with a whitelist | |
| 202 | - * of nodes and values. | |
| 203 | - * | |
| 204 | - * @var mixed[][] | |
| 205 | - */ | |
| 206 | - private $cssToHtmlMap = [ | |
| 207 | - 'background-color' => [ | |
| 208 | - 'attribute' => 'bgcolor', | |
| 209 | - ], | |
| 210 | - 'text-align' => [ | |
| 211 | - 'attribute' => 'align', | |
| 212 | - 'nodes' => ['p', 'div', 'td'], | |
| 213 | - 'values' => ['left', 'right', 'center', 'justify'], | |
| 214 | - ], | |
| 215 | - 'float' => [ | |
| 216 | - 'attribute' => 'align', | |
| 217 | - 'nodes' => ['table', 'img'], | |
| 218 | - 'values' => ['left', 'right'], | |
| 219 | - ], | |
| 220 | - 'border-spacing' => [ | |
| 221 | - 'attribute' => 'cellspacing', | |
| 222 | - 'nodes' => ['table'], | |
| 223 | - ], | |
| 224 | - ]; | |
| 225 | - | |
| 226 | - /** | |
| 227 | - * Emogrifier will throw Exceptions when it encounters an error instead of silently ignoring them. | |
| 228 | - * | |
| 229 | - * @var bool | |
| 230 | - */ | |
| 231 | - private $debug = false; | |
| 232 | - | |
| 233 | - /** | |
| 234 | - * The constructor. | |
| 235 | - * | |
| 236 | - * @param string $html the HTML to emogrify, must be UTF-8-encoded | |
| 237 | - * @param string $css the CSS to merge, must be UTF-8-encoded | |
| 238 | - */ | |
| 239 | - public function __construct( $html = '', $css = '' ) { | |
| 240 | - $this->setHtml( $html ); | |
| 241 | - $this->setCss( $css ); | |
| 242 | - } | |
| 243 | - | |
| 244 | - /** | |
| 245 | - * The destructor. | |
| 246 | - */ | |
| 247 | - public function __destruct() { | |
| 248 | - $this->purgeVisitedNodes(); | |
| 249 | - } | |
| 250 | - | |
| 251 | - /** | |
| 252 | - * Sets the HTML to emogrify. | |
| 253 | - * | |
| 254 | - * @param string $html the HTML to emogrify, must be UTF-8-encoded | |
| 255 | - * | |
| 256 | - * @return void | |
| 257 | - */ | |
| 258 | - public function setHtml( $html ) { | |
| 259 | - $this->html = $html; | |
| 260 | - } | |
| 261 | - | |
| 262 | - /** | |
| 263 | - * Sets the CSS to merge with the HTML. | |
| 264 | - * | |
| 265 | - * @param string $css the CSS to merge, must be UTF-8-encoded | |
| 266 | - * | |
| 267 | - * @return void | |
| 268 | - */ | |
| 269 | - public function setCss( $css ) { | |
| 270 | - $this->css = $css; | |
| 271 | - } | |
| 272 | - | |
| 273 | - /** | |
| 274 | - * Applies $this->css to $this->html and returns the HTML with the CSS | |
| 275 | - * applied. | |
| 276 | - * | |
| 277 | - * This method places the CSS inline. | |
| 278 | - * | |
| 279 | - * @return string | |
| 280 | - * | |
| 281 | - * @throws BadMethodCallException | |
| 282 | - */ | |
| 283 | - public function emogrify() { | |
| 284 | - if ( $this->html === '' ) { | |
| 285 | - throw new BadMethodCallException( 'Please set some HTML first before calling emogrify.', 1390393096 ); | |
| 286 | - } | |
| 287 | - | |
| 288 | - $xmlDocument = $this->createXmlDocument(); | |
| 289 | - $this->process( $xmlDocument ); | |
| 290 | - | |
| 291 | - return $xmlDocument->saveHTML(); | |
| 292 | - } | |
| 293 | - | |
| 294 | - /** | |
| 295 | - * Applies $this->css to $this->html and returns only the HTML content | |
| 296 | - * within the <body> tag. | |
| 297 | - * | |
| 298 | - * This method places the CSS inline. | |
| 299 | - * | |
| 300 | - * @return string | |
| 301 | - * | |
| 302 | - * @throws BadMethodCallException | |
| 303 | - */ | |
| 304 | - public function emogrifyBodyContent() { | |
| 305 | - if ( $this->html === '' ) { | |
| 306 | - throw new BadMethodCallException( 'Please set some HTML first before calling emogrify.', 1390393096 ); | |
| 307 | - } | |
| 308 | - | |
| 309 | - $xmlDocument = $this->createXmlDocument(); | |
| 310 | - $this->process( $xmlDocument ); | |
| 311 | - | |
| 312 | - $innerDocument = new \DOMDocument(); | |
| 313 | - | |
| 314 | - foreach ( $xmlDocument->documentElement->getElementsByTagName( 'body' )->item( 0 )->childNodes as $childNode ) { | |
| 315 | - $innerDocument->appendChild( $innerDocument->importNode( $childNode, true ) ); | |
| 316 | - } | |
| 317 | - | |
| 318 | - return html_entity_decode( $innerDocument->saveHTML() ); | |
| 319 | - } | |
| 320 | - | |
| 321 | - /** | |
| 322 | - * Applies $this->css to $xmlDocument. | |
| 323 | - * | |
| 324 | - * This method places the CSS inline. | |
| 325 | - * | |
| 326 | - * @param \DOMDocument $xmlDocument | |
| 327 | - * | |
| 328 | - * @return void | |
| 329 | - * | |
| 330 | - * @throws InvalidArgumentException | |
| 331 | - */ | |
| 332 | - protected function process( DOMDocument $xmlDocument ) { | |
| 333 | - $xPath = new \DOMXPath( $xmlDocument ); | |
| 334 | - $this->clearAllCaches(); | |
| 335 | - | |
| 336 | - // Before be begin processing the CSS file, parse the document and normalize all existing CSS attributes. | |
| 337 | - // This changes 'DISPLAY: none' to 'display: none'. | |
| 338 | - // We wouldn't have to do this if DOMXPath supported XPath 2.0. | |
| 339 | - // Also store a reference of nodes with existing inline styles so we don't overwrite them. | |
| 340 | - $this->purgeVisitedNodes(); | |
| 341 | - | |
| 342 | - set_error_handler( [$this, 'handleXpathError'], E_WARNING ); | |
| 343 | - | |
| 344 | - $nodesWithStyleAttributes = $xPath->query( '//*[@style]' ); | |
| 345 | - | |
| 346 | - if ( $nodesWithStyleAttributes !== false ) { | |
| 347 | - /** @var \DOMElement $node */ | |
| 348 | - foreach ( $nodesWithStyleAttributes as $node ) { | |
| 349 | - if ( $this->isInlineStyleAttributesParsingEnabled ) { | |
| 350 | - $this->normalizeStyleAttributes( $node ); | |
| 351 | - } else { | |
| 352 | - $node->removeAttribute( 'style' ); | |
| 353 | - } | |
| 354 | - } | |
| 355 | - } | |
| 356 | - | |
| 357 | - // grab any existing style blocks from the html and append them to the existing CSS | |
| 358 | - // (these blocks should be appended so as to have precedence over conflicting styles in the existing CSS) | |
| 359 | - $allCss = $this->css; | |
| 360 | - | |
| 361 | - if ( $this->isStyleBlocksParsingEnabled ) { | |
| 362 | - $allCss .= $this->getCssFromAllStyleNodes( $xPath ); | |
| 363 | - } | |
| 364 | - | |
| 365 | - $cssParts = $this->splitCssAndMediaQuery( $allCss ); | |
| 366 | - $excludedNodes = $this->getNodesToExclude( $xPath ); | |
| 367 | - $cssRules = $this->parseCssRules( $cssParts['css'] ); | |
| 368 | - | |
| 369 | - foreach ( $cssRules as $cssRule ) { | |
| 370 | - // query the body for the xpath selector | |
| 371 | - $nodesMatchingCssSelectors = $xPath->query( $this->translateCssToXpath( $cssRule['selector'] ) ); | |
| 372 | - // ignore invalid selectors | |
| 373 | - if ( $nodesMatchingCssSelectors === false ) { | |
| 374 | - continue; | |
| 375 | - } | |
| 376 | - | |
| 377 | - /** @var \DOMElement $node */ | |
| 378 | - foreach ( $nodesMatchingCssSelectors as $node ) { | |
| 379 | - if ( in_array( $node, $excludedNodes, true ) ) { | |
| 380 | - continue; | |
| 381 | - } | |
| 382 | - | |
| 383 | - // if it has a style attribute, get it, process it, and append (overwrite) new stuff | |
| 384 | - if ( $node->hasAttribute( 'style' ) ) { | |
| 385 | - // break it up into an associative array | |
| 386 | - $oldStyleDeclarations = $this->parseCssDeclarationsBlock( $node->getAttribute( 'style' ) ); | |
| 387 | - } else { | |
| 388 | - $oldStyleDeclarations = []; | |
| 389 | - } | |
| 390 | - $newStyleDeclarations = $this->parseCssDeclarationsBlock( $cssRule['declarationsBlock'] ); | |
| 391 | - | |
| 392 | - if ( $this->shouldMapCssToHtml ) { | |
| 393 | - $this->mapCssToHtmlAttributes( $newStyleDeclarations, $node ); | |
| 394 | - } | |
| 395 | - $node->setAttribute( | |
| 396 | - 'style', | |
| 397 | - $this->generateStyleStringFromDeclarationsArrays( $oldStyleDeclarations, $newStyleDeclarations ) | |
| 398 | - ); | |
| 399 | - } | |
| 400 | - } | |
| 401 | - | |
| 402 | - restore_error_handler(); | |
| 403 | - | |
| 404 | - if ( $this->isInlineStyleAttributesParsingEnabled ) { | |
| 405 | - $this->fillStyleAttributesWithMergedStyles(); | |
| 406 | - } | |
| 407 | - | |
| 408 | - if ( $this->shouldKeepInvisibleNodes ) { | |
| 409 | - $this->removeInvisibleNodes( $xPath ); | |
| 410 | - } | |
| 411 | - | |
| 412 | - $this->copyCssWithMediaToStyleNode( $xmlDocument, $xPath, $cssParts['media'] ); | |
| 413 | - } | |
| 414 | - | |
| 415 | - /** | |
| 416 | - * Applies $styles to $node. | |
| 417 | - * | |
| 418 | - * This method maps CSS styles to HTML attributes and adds those to the | |
| 419 | - * node. | |
| 420 | - * | |
| 421 | - * @param string[] $styles the new CSS styles taken from the global styles to be applied to this node | |
| 422 | - * @param \DOMNode $node node to apply styles to | |
| 423 | - * | |
| 424 | - * @return void | |
| 425 | - */ | |
| 426 | - private function mapCssToHtmlAttributes( array $styles, DOMNode $node ) { | |
| 427 | - foreach ( $styles as $property => $value ) { | |
| 428 | - // Strip !important indicator | |
| 429 | - $value = trim( str_replace( '!important', '', $value ) ); | |
| 430 | - $this->mapCssToHtmlAttribute( $property, $value, $node ); | |
| 431 | - } | |
| 432 | - } | |
| 433 | - | |
| 434 | - /** | |
| 435 | - * Tries to apply the CSS style to $node as an attribute. | |
| 436 | - * | |
| 437 | - * This method maps a CSS rule to HTML attributes and adds those to the node. | |
| 438 | - * | |
| 439 | - * @param string $property the name of the CSS property to map | |
| 440 | - * @param string $value the value of the style rule to map | |
| 441 | - * @param \DOMNode $node node to apply styles to | |
| 442 | - * | |
| 443 | - * @return void | |
| 444 | - */ | |
| 445 | - private function mapCssToHtmlAttribute( $property, $value, DOMNode $node ) { | |
| 446 | - if ( !$this->mapSimpleCssProperty( $property, $value, $node ) ) { | |
| 447 | - $this->mapComplexCssProperty( $property, $value, $node ); | |
| 448 | - } | |
| 449 | - } | |
| 450 | - | |
| 451 | - /** | |
| 452 | - * Looks up the CSS property in the mapping table and maps it if it matches the conditions. | |
| 453 | - * | |
| 454 | - * @param string $property the name of the CSS property to map | |
| 455 | - * @param string $value the value of the style rule to map | |
| 456 | - * @param \DOMNode $node node to apply styles to | |
| 457 | - * | |
| 458 | - * @return bool true if the property cab be mapped using the simple mapping table | |
| 459 | - */ | |
| 460 | - private function mapSimpleCssProperty( $property, $value, DOMNode $node ) { | |
| 461 | - if ( !isset( $this->cssToHtmlMap[$property] ) ) { | |
| 462 | - return false; | |
| 463 | - } | |
| 464 | - | |
| 465 | - $mapping = $this->cssToHtmlMap[$property]; | |
| 466 | - $nodesMatch = !isset( $mapping['nodes'] ) || in_array( $node->nodeName, $mapping['nodes'], true ); | |
| 467 | - $valuesMatch = !isset( $mapping['values'] ) || in_array( $value, $mapping['values'], true ); | |
| 468 | - | |
| 469 | - if ( !$nodesMatch || !$valuesMatch ) { | |
| 470 | - return false; | |
| 471 | - } | |
| 472 | - | |
| 473 | - $node->setAttribute( $mapping['attribute'], $value ); | |
| 474 | - | |
| 475 | - return true; | |
| 476 | - } | |
| 477 | - | |
| 478 | - /** | |
| 479 | - * Maps CSS properties that need special transformation to an HTML attribute. | |
| 480 | - * | |
| 481 | - * @param string $property the name of the CSS property to map | |
| 482 | - * @param string $value the value of the style rule to map | |
| 483 | - * @param \DOMNode $node node to apply styles to | |
| 484 | - * | |
| 485 | - * @return void | |
| 486 | - */ | |
| 487 | - private function mapComplexCssProperty( $property, $value, DOMNode $node ) { | |
| 488 | - $nodeName = $node->nodeName; | |
| 489 | - $isTable = $nodeName === 'table'; | |
| 490 | - $isImage = $nodeName === 'img'; | |
| 491 | - $isTableOrImage = $isTable || $isImage; | |
| 492 | - | |
| 493 | - switch ( $property ) { | |
| 494 | - case 'background': | |
| 495 | - // Parse out the color, if any | |
| 496 | - $styles = explode( ' ', $value ); | |
| 497 | - $first = $styles[0]; | |
| 498 | - | |
| 499 | - if ( !is_numeric( substr( $first, 0, 1 ) ) && substr( $first, 0, 3 ) !== 'url' ) { | |
| 500 | - // This is not a position or image, assume it's a color | |
| 501 | - $node->setAttribute( 'bgcolor', $first ); | |
| 502 | - } | |
| 503 | - break; | |
| 504 | - | |
| 505 | - case 'width': | |
| 506 | - // intentional fall-through | |
| 507 | - case 'height': | |
| 508 | - // Only parse values in px and %, but not values like "auto". | |
| 509 | - if ( preg_match( '/^\d+(px|%)$/', $value ) ) { | |
| 510 | - // Remove 'px'. This regex only conserves numbers and % | |
| 511 | - $number = preg_replace( '/[^0-9.%]/', '', $value ); | |
| 512 | - $node->setAttribute( $property, $number ); | |
| 513 | - } | |
| 514 | - break; | |
| 515 | - | |
| 516 | - case 'margin': | |
| 517 | - if ( $isTableOrImage ) { | |
| 518 | - $margins = $this->parseCssShorthandValue( $value ); | |
| 519 | - | |
| 520 | - if ( $margins['left'] === 'auto' && $margins['right'] === 'auto' ) { | |
| 521 | - $node->setAttribute( 'align', 'center' ); | |
| 522 | - } | |
| 523 | - } | |
| 524 | - break; | |
| 525 | - | |
| 526 | - case 'border': | |
| 527 | - if ( $isTableOrImage ) { | |
| 528 | - if ( $value === 'none' || $value === '0' ) { | |
| 529 | - $node->setAttribute( 'border', '0' ); | |
| 530 | - } | |
| 531 | - } | |
| 532 | - break; | |
| 533 | - default: | |
| 534 | - } | |
| 535 | - } | |
| 536 | - | |
| 537 | - /** | |
| 538 | - * Parses a shorthand CSS value and splits it into individual values | |
| 539 | - * | |
| 540 | - * @param string $value a string of CSS value with 1, 2, 3 or 4 sizes | |
| 541 | - * For example: padding: 0 auto; | |
| 542 | - * '0 auto' is split into top: 0, left: auto, bottom: 0, | |
| 543 | - * right: auto | |
| 544 | - * | |
| 545 | - * @return string[] an array of values for top, right, bottom and left (using these as associative array keys) | |
| 546 | - */ | |
| 547 | - private function parseCssShorthandValue( $value ) { | |
| 548 | - $values = preg_split( '/\\s+/', $value ); | |
| 549 | - | |
| 550 | - $css = []; | |
| 551 | - $css['top'] = $values[0]; | |
| 552 | - $css['right'] = ( count( $values ) > 1 ) ? $values[1] : $css['top']; | |
| 553 | - $css['bottom'] = ( count( $values ) > 2 ) ? $values[2] : $css['top']; | |
| 554 | - $css['left'] = ( count( $values ) > 3 ) ? $values[3] : $css['right']; | |
| 555 | - | |
| 556 | - return $css; | |
| 557 | - } | |
| 558 | - | |
| 559 | - /** | |
| 560 | - * Extracts and parses the individual rules from a CSS string. | |
| 561 | - * | |
| 562 | - * @param string $css a string of raw CSS code | |
| 563 | - * | |
| 564 | - * @return string[][] an array of string sub-arrays with the keys | |
| 565 | - * "selector" (the CSS selector(s), e.g., "*" or "h1"), | |
| 566 | - * "declarationsBLock" (the semicolon-separated CSS declarations for that selector(s), | |
| 567 | - * e.g., "color: red; height: 4px;"), | |
| 568 | - * and "line" (the line number e.g. 42) | |
| 569 | - */ | |
| 570 | - private function parseCssRules( $css ) { | |
| 571 | - $cssKey = md5( $css ); | |
| 572 | - | |
| 573 | - if ( !isset( $this->caches[self::CACHE_KEY_CSS][$cssKey] ) ) { | |
| 574 | - // process the CSS file for selectors and definitions | |
| 575 | - preg_match_all( '/(?:^|[\\s^{}]*)([^{]+){([^}]*)}/mis', $css, $matches, PREG_SET_ORDER ); | |
| 576 | - | |
| 577 | - $cssRules = []; | |
| 578 | - /** @var string[] $cssRule */ | |
| 579 | - foreach ( $matches as $key => $cssRule ) { | |
| 580 | - $cssDeclaration = trim( $cssRule[2] ); | |
| 581 | - | |
| 582 | - if ( $cssDeclaration === '' ) { | |
| 583 | - continue; | |
| 584 | - } | |
| 585 | - | |
| 586 | - $selectors = explode( ',', $cssRule[1] ); | |
| 587 | - | |
| 588 | - foreach ( $selectors as $selector ) { | |
| 589 | - // don't process pseudo-elements and behavioral (dynamic) pseudo-classes; | |
| 590 | - // only allow structural pseudo-classes | |
| 591 | - $hasPseudoElement = strpos( $selector, '::' ) !== false; | |
| 592 | - $hasAnyPseudoClass = (bool) preg_match( '/:[a-zA-Z]/', $selector ); | |
| 593 | - $hasSupportedPseudoClass = (bool) preg_match( '/:\\S+\\-(child|type\\()/i', $selector ); | |
| 594 | - | |
| 595 | - if ( $hasPseudoElement || ( $hasAnyPseudoClass && !$hasSupportedPseudoClass ) ) { | |
| 596 | - continue; | |
| 597 | - } | |
| 598 | - | |
| 599 | - $cssRules[] = [ | |
| 600 | - 'selector' => trim( $selector ), | |
| 601 | - 'declarationsBlock' => $cssDeclaration, | |
| 602 | - // keep track of where it appears in the file, since order is important | |
| 603 | - 'line' => $key, | |
| 604 | - ]; | |
| 605 | - } | |
| 606 | - } | |
| 607 | - | |
| 608 | - usort( $cssRules, [$this, 'sortBySelectorPrecedence'] ); | |
| 609 | - | |
| 610 | - $this->caches[self::CACHE_KEY_CSS][$cssKey] = $cssRules; | |
| 611 | - } | |
| 612 | - | |
| 613 | - return $this->caches[self::CACHE_KEY_CSS][$cssKey]; | |
| 614 | - } | |
| 615 | - | |
| 616 | - /** | |
| 617 | - * Disables the parsing of inline styles. | |
| 618 | - * | |
| 619 | - * @return void | |
| 620 | - */ | |
| 621 | - public function disableInlineStyleAttributesParsing() { | |
| 622 | - $this->isInlineStyleAttributesParsingEnabled = false; | |
| 623 | - } | |
| 624 | - | |
| 625 | - /** | |
| 626 | - * Disables the parsing of <style> blocks. | |
| 627 | - * | |
| 628 | - * @return void | |
| 629 | - */ | |
| 630 | - public function disableStyleBlocksParsing() { | |
| 631 | - $this->isStyleBlocksParsingEnabled = false; | |
| 632 | - } | |
| 633 | - | |
| 634 | - /** | |
| 635 | - * Disables the removal of elements with `display: none` properties. | |
| 636 | - * | |
| 637 | - * @return void | |
| 638 | - */ | |
| 639 | - public function disableInvisibleNodeRemoval() { | |
| 640 | - $this->shouldKeepInvisibleNodes = false; | |
| 641 | - } | |
| 642 | - | |
| 643 | - /** | |
| 644 | - * Enables the attachment/override of HTML attributes for which a | |
| 645 | - * corresponding CSS property has been set. | |
| 646 | - * | |
| 647 | - * @return void | |
| 648 | - */ | |
| 649 | - public function enableCssToHtmlMapping() { | |
| 650 | - $this->shouldMapCssToHtml = true; | |
| 651 | - } | |
| 652 | - | |
| 653 | - /** | |
| 654 | - * Clears all caches. | |
| 655 | - * | |
| 656 | - * @return void | |
| 657 | - */ | |
| 658 | - private function clearAllCaches() { | |
| 659 | - $this->clearCache( self::CACHE_KEY_CSS ); | |
| 660 | - $this->clearCache( self::CACHE_KEY_SELECTOR ); | |
| 661 | - $this->clearCache( self::CACHE_KEY_XPATH ); | |
| 662 | - $this->clearCache( self::CACHE_KEY_CSS_DECLARATIONS_BLOCK ); | |
| 663 | - $this->clearCache( self::CACHE_KEY_COMBINED_STYLES ); | |
| 664 | - } | |
| 665 | - | |
| 666 | - /** | |
| 667 | - * Clears a single cache by key. | |
| 668 | - * | |
| 669 | - * @param int $key the cache key, must be CACHE_KEY_CSS, CACHE_KEY_SELECTOR, CACHE_KEY_XPATH | |
| 670 | - * or CACHE_KEY_CSS_DECLARATION_BLOCK | |
| 671 | - * | |
| 672 | - * @return void | |
| 673 | - * | |
| 674 | - * @throws InvalidArgumentException | |
| 675 | - */ | |
| 676 | - private function clearCache( $key ) { | |
| 677 | - $allowedCacheKeys = [ | |
| 678 | - self::CACHE_KEY_CSS, | |
| 679 | - self::CACHE_KEY_SELECTOR, | |
| 680 | - self::CACHE_KEY_XPATH, | |
| 681 | - self::CACHE_KEY_CSS_DECLARATIONS_BLOCK, | |
| 682 | - self::CACHE_KEY_COMBINED_STYLES, | |
| 683 | - ]; | |
| 684 | - | |
| 685 | - if ( !in_array( $key, $allowedCacheKeys, true ) ) { | |
| 686 | - throw new InvalidArgumentException( 'Invalid cache key: ' . $key, 1391822035 ); | |
| 687 | - } | |
| 688 | - | |
| 689 | - $this->caches[$key] = []; | |
| 690 | - } | |
| 691 | - | |
| 692 | - /** | |
| 693 | - * Purges the visited nodes. | |
| 694 | - * | |
| 695 | - * @return void | |
| 696 | - */ | |
| 697 | - private function purgeVisitedNodes() { | |
| 698 | - $this->visitedNodes = []; | |
| 699 | - $this->styleAttributesForNodes = []; | |
| 700 | - } | |
| 701 | - | |
| 702 | - /** | |
| 703 | - * Marks a tag for removal. | |
| 704 | - * | |
| 705 | - * There are some HTML tags that DOMDocument cannot process, and it will throw an error if it encounters them. | |
| 706 | - * In particular, DOMDocument will complain if you try to use HTML5 tags in an XHTML document. | |
| 707 | - * | |
| 708 | - * Note: The tags will not be removed if they have any content. | |
| 709 | - * | |
| 710 | - * @param string $tagName the tag name, e.g., "p" | |
| 711 | - * | |
| 712 | - * @return void | |
| 713 | - */ | |
| 714 | - public function addUnprocessableHtmlTag( $tagName ) { | |
| 715 | - $this->unprocessableHtmlTags[] = $tagName; | |
| 716 | - } | |
| 717 | - | |
| 718 | - /** | |
| 719 | - * Drops a tag from the removal list. | |
| 720 | - * | |
| 721 | - * @param string $tagName the tag name, e.g., "p" | |
| 722 | - * | |
| 723 | - * @return void | |
| 724 | - */ | |
| 725 | - public function removeUnprocessableHtmlTag( $tagName ) { | |
| 726 | - $key = array_search( $tagName, $this->unprocessableHtmlTags, true ); | |
| 727 | - | |
| 728 | - if ( $key !== false ) { | |
| 729 | - unset( $this->unprocessableHtmlTags[$key] ); | |
| 730 | - } | |
| 731 | - } | |
| 732 | - | |
| 733 | - /** | |
| 734 | - * Marks a media query type to keep. | |
| 735 | - * | |
| 736 | - * @param string $mediaName the media type name, e.g., "braille" | |
| 737 | - * | |
| 738 | - * @return void | |
| 739 | - */ | |
| 740 | - public function addAllowedMediaType( $mediaName ) { | |
| 741 | - $this->allowedMediaTypes[$mediaName] = true; | |
| 742 | - } | |
| 743 | - | |
| 744 | - /** | |
| 745 | - * Drops a media query type from the allowed list. | |
| 746 | - * | |
| 747 | - * @param string $mediaName the tag name, e.g., "braille" | |
| 748 | - * | |
| 749 | - * @return void | |
| 750 | - */ | |
| 751 | - public function removeAllowedMediaType( $mediaName ) { | |
| 752 | - if ( isset( $this->allowedMediaTypes[$mediaName] ) ) { | |
| 753 | - unset( $this->allowedMediaTypes[$mediaName] ); | |
| 754 | - } | |
| 755 | - } | |
| 756 | - | |
| 757 | - /** | |
| 758 | - * Adds a selector to exclude nodes from emogrification. | |
| 759 | - * | |
| 760 | - * Any nodes that match the selector will not have their style altered. | |
| 761 | - * | |
| 762 | - * @param string $selector the selector to exclude, e.g., ".editor" | |
| 763 | - * | |
| 764 | - * @return void | |
| 765 | - */ | |
| 766 | - public function addExcludedSelector( $selector ) { | |
| 767 | - $this->excludedSelectors[$selector] = true; | |
| 768 | - } | |
| 769 | - | |
| 770 | - /** | |
| 771 | - * No longer excludes the nodes matching this selector from emogrification. | |
| 772 | - * | |
| 773 | - * @param string $selector the selector to no longer exclude, e.g., ".editor" | |
| 774 | - * | |
| 775 | - * @return void | |
| 776 | - */ | |
| 777 | - public function removeExcludedSelector( $selector ) { | |
| 778 | - if ( isset( $this->excludedSelectors[$selector] ) ) { | |
| 779 | - unset( $this->excludedSelectors[$selector] ); | |
| 780 | - } | |
| 781 | - } | |
| 782 | - | |
| 783 | - /** | |
| 784 | - * This removes styles from your email that contain display:none. | |
| 785 | - * We need to look for display:none, but we need to do a case-insensitive search. Since DOMDocument only | |
| 786 | - * supports XPath 1.0, lower-case() isn't available to us. We've thus far only set attributes to lowercase, | |
| 787 | - * not attribute values. Consequently, we need to translate() the letters that would be in 'NONE' ("NOE") | |
| 788 | - * to lowercase. | |
| 789 | - * | |
| 790 | - * @param \DOMXPath $xPath | |
| 791 | - * | |
| 792 | - * @return void | |
| 793 | - */ | |
| 794 | - private function removeInvisibleNodes( DOMXPath $xPath ) { | |
| 795 | - $nodesWithStyleDisplayNone = $xPath->query( | |
| 796 | - '//*[contains(translate(translate(@style," ",""),"NOE","noe"),"display:none")]' | |
| 797 | - ); | |
| 798 | - | |
| 799 | - if ( $nodesWithStyleDisplayNone->length === 0 ) { | |
| 800 | - return; | |
| 801 | - } | |
| 802 | - | |
| 803 | - // The checks on parentNode and is_callable below ensure that if we've deleted the parent node, | |
| 804 | - // we don't try to call removeChild on a nonexistent child node | |
| 805 | - /** @var \DOMNode $node */ | |
| 806 | - foreach ( $nodesWithStyleDisplayNone as $node ) { | |
| 807 | - if ( $node->parentNode && is_callable( [$node->parentNode, 'removeChild'] ) ) { | |
| 808 | - $node->parentNode->removeChild( $node ); | |
| 809 | - } | |
| 810 | - } | |
| 811 | - } | |
| 812 | - | |
| 813 | - /** | |
| 814 | - * Normalizes the value of the "style" attribute and saves it. | |
| 815 | - * | |
| 816 | - * @param \DOMElement $node | |
| 817 | - * | |
| 818 | - * @return void | |
| 819 | - */ | |
| 820 | - private function normalizeStyleAttributes( DOMElement $node ) { | |
| 821 | - $normalizedOriginalStyle = preg_replace_callback( | |
| 822 | - '/[A-z\\-]+(?=\\:)/S', | |
| 823 | - function ( array $m ) { | |
| 824 | - return strtolower( $m[0] ); | |
| 825 | - }, | |
| 826 | - $node->getAttribute( 'style' ) | |
| 827 | - ); | |
| 828 | - | |
| 829 | - // in order to not overwrite existing style attributes in the HTML, we | |
| 830 | - // have to save the original HTML styles | |
| 831 | - $nodePath = $node->getNodePath(); | |
| 832 | - | |
| 833 | - if ( !isset( $this->styleAttributesForNodes[$nodePath] ) ) { | |
| 834 | - $this->styleAttributesForNodes[$nodePath] = $this->parseCssDeclarationsBlock( $normalizedOriginalStyle ); | |
| 835 | - $this->visitedNodes[$nodePath] = $node; | |
| 836 | - } | |
| 837 | - | |
| 838 | - $node->setAttribute( 'style', $normalizedOriginalStyle ); | |
| 839 | - } | |
| 840 | - | |
| 841 | - /** | |
| 842 | - * Merges styles from styles attributes and style nodes and applies them to the attribute nodes | |
| 843 | - * | |
| 844 | - * @return void | |
| 845 | - */ | |
| 846 | - private function fillStyleAttributesWithMergedStyles() { | |
| 847 | - foreach ( $this->styleAttributesForNodes as $nodePath => $styleAttributesForNode ) { | |
| 848 | - $node = $this->visitedNodes[$nodePath]; | |
| 849 | - $currentStyleAttributes = $this->parseCssDeclarationsBlock( $node->getAttribute( 'style' ) ); | |
| 850 | - $node->setAttribute( | |
| 851 | - 'style', | |
| 852 | - $this->generateStyleStringFromDeclarationsArrays( | |
| 853 | - $currentStyleAttributes, | |
| 854 | - $styleAttributesForNode | |
| 855 | - ) | |
| 856 | - ); | |
| 857 | - } | |
| 858 | - } | |
| 859 | - | |
| 860 | - /** | |
| 861 | - * This method merges old or existing name/value array with new name/value array | |
| 862 | - * and then generates a string of the combined style suitable for placing inline. | |
| 863 | - * This becomes the single point for CSS string generation allowing for consistent | |
| 864 | - * CSS output no matter where the CSS originally came from. | |
| 865 | - * | |
| 866 | - * @param string[] $oldStyles | |
| 867 | - * @param string[] $newStyles | |
| 868 | - * | |
| 869 | - * @return string | |
| 870 | - */ | |
| 871 | - private function generateStyleStringFromDeclarationsArrays( array $oldStyles, array $newStyles ) { | |
| 872 | - $combinedStyles = array_merge( $oldStyles, $newStyles ); | |
| 873 | - $cacheKey = serialize( $combinedStyles ); | |
| 874 | - | |
| 875 | - if ( isset( $this->caches[self::CACHE_KEY_COMBINED_STYLES][$cacheKey] ) ) { | |
| 876 | - return $this->caches[self::CACHE_KEY_COMBINED_STYLES][$cacheKey]; | |
| 877 | - } | |
| 878 | - | |
| 879 | - foreach ( $oldStyles as $attributeName => $attributeValue ) { | |
| 880 | - if ( !isset( $newStyles[$attributeName] ) ) { | |
| 881 | - continue; | |
| 882 | - } | |
| 883 | - | |
| 884 | - $newAttributeValue = $newStyles[$attributeName]; | |
| 885 | - | |
| 886 | - if ( $this->attributeValueIsImportant( $attributeValue ) | |
| 887 | - && !$this->attributeValueIsImportant( $newAttributeValue ) | |
| 888 | - ) { | |
| 889 | - $combinedStyles[$attributeName] = $attributeValue; | |
| 890 | - } | |
| 891 | - } | |
| 892 | - | |
| 893 | - $style = ''; | |
| 894 | - | |
| 895 | - foreach ( $combinedStyles as $attributeName => $attributeValue ) { | |
| 896 | - $style .= strtolower( trim( $attributeName ) ) . ': ' . trim( $attributeValue ) . '; '; | |
| 897 | - } | |
| 898 | - $trimmedStyle = rtrim( $style ); | |
| 899 | - | |
| 900 | - $this->caches[self::CACHE_KEY_COMBINED_STYLES][$cacheKey] = $trimmedStyle; | |
| 901 | - | |
| 902 | - return $trimmedStyle; | |
| 903 | - } | |
| 904 | - | |
| 905 | - /** | |
| 906 | - * Checks whether $attributeValue is marked as !important. | |
| 907 | - * | |
| 908 | - * @param string $attributeValue | |
| 909 | - * | |
| 910 | - * @return bool | |
| 911 | - */ | |
| 912 | - private function attributeValueIsImportant( $attributeValue ) { | |
| 913 | - return strtolower( substr( trim( $attributeValue ), -10 ) ) === '!important'; | |
| 914 | - } | |
| 915 | - | |
| 916 | - /** | |
| 917 | - * Applies $css to $xmlDocument, limited to the media queries that actually apply to the document. | |
| 918 | - * | |
| 919 | - * @param \DOMDocument $xmlDocument the document to match against | |
| 920 | - * @param \DOMXPath $xPath | |
| 921 | - * @param string $css a string of CSS | |
| 922 | - * | |
| 923 | - * @return void | |
| 924 | - */ | |
| 925 | - private function copyCssWithMediaToStyleNode( DOMDocument $xmlDocument, DOMXPath $xPath, $css ) { | |
| 926 | - if ( $css === '' ) { | |
| 927 | - return; | |
| 928 | - } | |
| 929 | - | |
| 930 | - $mediaQueriesRelevantForDocument = []; | |
| 931 | - | |
| 932 | - foreach ( $this->extractMediaQueriesFromCss( $css ) as $mediaQuery ) { | |
| 933 | - foreach ( $this->parseCssRules( $mediaQuery['css'] ) as $selector ) { | |
| 934 | - if ( $this->existsMatchForCssSelector( $xPath, $selector['selector'] ) ) { | |
| 935 | - $mediaQueriesRelevantForDocument[] = $mediaQuery['query']; | |
| 936 | - break; | |
| 937 | - } | |
| 938 | - } | |
| 939 | - } | |
| 940 | - | |
| 941 | - $this->addStyleElementToDocument( $xmlDocument, implode( $mediaQueriesRelevantForDocument ) ); | |
| 942 | - } | |
| 943 | - | |
| 944 | - /** | |
| 945 | - * Extracts the media queries from $css while skipping empty media queries. | |
| 946 | - * | |
| 947 | - * @param string $css | |
| 948 | - * | |
| 949 | - * @return string[][] numeric array with string sub-arrays with the keys "css" and "query" | |
| 950 | - */ | |
| 951 | - private function extractMediaQueriesFromCss( $css ) { | |
| 952 | - preg_match_all( '/@media\\b[^{]*({((?:[^{}]+|(?1))*)})/', $css, $rawMediaQueries, PREG_SET_ORDER ); | |
| 953 | - $parsedQueries = []; | |
| 954 | - | |
| 955 | - foreach ( $rawMediaQueries as $mediaQuery ) { | |
| 956 | - if ( $mediaQuery[2] !== '' ) { | |
| 957 | - $parsedQueries[] = [ | |
| 958 | - 'css' => $mediaQuery[2], | |
| 959 | - 'query' => $mediaQuery[0], | |
| 960 | - ]; | |
| 961 | - } | |
| 962 | - } | |
| 963 | - | |
| 964 | - return $parsedQueries; | |
| 965 | - } | |
| 966 | - | |
| 967 | - /** | |
| 968 | - * Checks whether there is at least one matching element for $cssSelector. | |
| 969 | - * | |
| 970 | - * @param \DOMXPath $xPath | |
| 971 | - * @param string $cssSelector | |
| 972 | - * | |
| 973 | - * @return bool | |
| 974 | - */ | |
| 975 | - private function existsMatchForCssSelector( DOMXPath $xPath, $cssSelector ) { | |
| 976 | - $nodesMatchingSelector = $xPath->query( $this->translateCssToXpath( $cssSelector ) ); | |
| 977 | - | |
| 978 | - return $nodesMatchingSelector !== false && $nodesMatchingSelector->length !== 0; | |
| 979 | - } | |
| 980 | - | |
| 981 | - /** | |
| 982 | - * Returns CSS content. | |
| 983 | - * | |
| 984 | - * @param \DOMXPath $xPath | |
| 985 | - * | |
| 986 | - * @return string | |
| 987 | - */ | |
| 988 | - private function getCssFromAllStyleNodes( DOMXPath $xPath ) { | |
| 989 | - $styleNodes = $xPath->query( '//style' ); | |
| 990 | - | |
| 991 | - if ( $styleNodes === false ) { | |
| 992 | - return ''; | |
| 993 | - } | |
| 994 | - | |
| 995 | - $css = ''; | |
| 996 | - /** @var \DOMNode $styleNode */ | |
| 997 | - foreach ( $styleNodes as $styleNode ) { | |
| 998 | - $css .= "\n\n" . $styleNode->nodeValue; | |
| 999 | - $styleNode->parentNode->removeChild( $styleNode ); | |
| 1000 | - } | |
| 1001 | - | |
| 1002 | - return $css; | |
| 1003 | - } | |
| 1004 | - | |
| 1005 | - /** | |
| 1006 | - * Adds a style element with $css to $document. | |
| 1007 | - * | |
| 1008 | - * This method is protected to allow overriding. | |
| 1009 | - * | |
| 1010 | - * @see https://github.com/jjriv/emogrifier/issues/103 | |
| 1011 | - * | |
| 1012 | - * @param \DOMDocument $document | |
| 1013 | - * @param string $css | |
| 1014 | - * | |
| 1015 | - * @return void | |
| 1016 | - */ | |
| 1017 | - protected function addStyleElementToDocument( DOMDocument $document, $css ) { | |
| 1018 | - $styleElement = $document->createElement( 'style', $css ); | |
| 1019 | - $styleAttribute = $document->createAttribute( 'type' ); | |
| 1020 | - $styleAttribute->value = 'text/css'; | |
| 1021 | - $styleElement->appendChild( $styleAttribute ); | |
| 1022 | - | |
| 1023 | - $head = $this->getOrCreateHeadElement( $document ); | |
| 1024 | - $head->appendChild( $styleElement ); | |
| 1025 | - } | |
| 1026 | - | |
| 1027 | - /** | |
| 1028 | - * Returns the existing or creates a new head element in $document. | |
| 1029 | - * | |
| 1030 | - * @param \DOMDocument $document | |
| 1031 | - * | |
| 1032 | - * @return \DOMNode the head element | |
| 1033 | - */ | |
| 1034 | - private function getOrCreateHeadElement( DOMDocument $document ) { | |
| 1035 | - $head = $document->getElementsByTagName( 'head' )->item( 0 ); | |
| 1036 | - | |
| 1037 | - if ( $head === null ) { | |
| 1038 | - $head = $document->createElement( 'head' ); | |
| 1039 | - $html = $document->getElementsByTagName( 'html' )->item( 0 ); | |
| 1040 | - $html->insertBefore( $head, $document->getElementsByTagName( 'body' )->item( 0 ) ); | |
| 1041 | - } | |
| 1042 | - | |
| 1043 | - return $head; | |
| 1044 | - } | |
| 1045 | - | |
| 1046 | - /** | |
| 1047 | - * Splits input CSS code to an array where: | |
| 1048 | - * | |
| 1049 | - * - key "css" will be contains clean CSS code | |
| 1050 | - * - key "media" will be contains all valuable media queries | |
| 1051 | - * | |
| 1052 | - * Example: | |
| 1053 | - * | |
| 1054 | - * The CSS code | |
| 1055 | - * | |
| 1056 | - * "@import "file.css"; h1 { color:red; } @media { h1 {}} @media tv { h1 {}}" | |
| 1057 | - * | |
| 1058 | - * will be parsed into the following array: | |
| 1059 | - * | |
| 1060 | - * "css" => "h1 { color:red; }" | |
| 1061 | - * "media" => "@media { h1 {}}" | |
| 1062 | - * | |
| 1063 | - * @param string $css | |
| 1064 | - * | |
| 1065 | - * @return string[] | |
| 1066 | - */ | |
| 1067 | - private function splitCssAndMediaQuery( $css ) { | |
| 1068 | - $cssWithoutComments = preg_replace( '/\\/\\*.*\\*\\//sU', '', $css ); | |
| 1069 | - | |
| 1070 | - $mediaTypesExpression = ''; | |
| 1071 | - | |
| 1072 | - if ( !empty( $this->allowedMediaTypes ) ) { | |
| 1073 | - $mediaTypesExpression = '|' . implode( '|', array_keys( $this->allowedMediaTypes ) ); | |
| 1074 | - } | |
| 1075 | - | |
| 1076 | - $media = ''; | |
| 1077 | - $cssForAllowedMediaTypes = preg_replace_callback( | |
| 1078 | - '#@media\\s+(?:only\\s)?(?:[\\s{\\(]' . $mediaTypesExpression . ')\\s?[^{]+{.*}\\s*}\\s*#misU', | |
| 1079 | - function ( $matches ) use ( &$media ) { | |
| 1080 | - $media .= $matches[0]; | |
| 1081 | - }, | |
| 1082 | - $cssWithoutComments | |
| 1083 | - ); | |
| 1084 | - | |
| 1085 | - // filter the CSS | |
| 1086 | - $search = [ | |
| 1087 | - 'import directives' => '/^\\s*@import\\s[^;]+;/misU', | |
| 1088 | - 'remaining media enclosures' => '/^\\s*@media\\s[^{]+{(.*)}\\s*}\\s/misU', | |
| 1089 | - ]; | |
| 1090 | - | |
| 1091 | - $cleanedCss = preg_replace( $search, '', $cssForAllowedMediaTypes ); | |
| 1092 | - | |
| 1093 | - return ['css' => $cleanedCss, 'media' => $media]; | |
| 1094 | - } | |
| 1095 | - | |
| 1096 | - /** | |
| 1097 | - * Creates a DOMDocument instance with the current HTML. | |
| 1098 | - * | |
| 1099 | - * @return \DOMDocument | |
| 1100 | - */ | |
| 1101 | - private function createXmlDocument() { | |
| 1102 | - $xmlDocument = new \DOMDocument(); | |
| 1103 | - $xmlDocument->encoding = 'UTF-8'; | |
| 1104 | - $xmlDocument->strictErrorChecking = false; | |
| 1105 | - $xmlDocument->formatOutput = true; | |
| 1106 | - $libXmlState = libxml_use_internal_errors( true ); | |
| 1107 | - $xmlDocument->loadHTML( $this->getUnifiedHtml() ); | |
| 1108 | - libxml_clear_errors(); | |
| 1109 | - libxml_use_internal_errors( $libXmlState ); | |
| 1110 | - $xmlDocument->normalizeDocument(); | |
| 1111 | - | |
| 1112 | - return $xmlDocument; | |
| 1113 | - } | |
| 1114 | - | |
| 1115 | - /** | |
| 1116 | - * Returns the HTML with the unprocessable HTML tags removed and | |
| 1117 | - * with added document type and Content-Type meta tag if needed. | |
| 1118 | - * | |
| 1119 | - * @return string the unified HTML | |
| 1120 | - * | |
| 1121 | - * @throws BadMethodCallException | |
| 1122 | - */ | |
| 1123 | - private function getUnifiedHtml() { | |
| 1124 | - $htmlWithoutUnprocessableTags = $this->removeUnprocessableTags( $this->html ); | |
| 1125 | - $htmlWithDocumentType = $this->ensureDocumentType( $htmlWithoutUnprocessableTags ); | |
| 1126 | - | |
| 1127 | - return $this->addContentTypeMetaTag( $htmlWithDocumentType ); | |
| 1128 | - } | |
| 1129 | - | |
| 1130 | - /** | |
| 1131 | - * Removes the unprocessable tags from $html (if this feature is enabled). | |
| 1132 | - * | |
| 1133 | - * @param string $html | |
| 1134 | - * | |
| 1135 | - * @return string the reworked HTML with the unprocessable tags removed | |
| 1136 | - */ | |
| 1137 | - private function removeUnprocessableTags( $html ) { | |
| 1138 | - if ( empty( $this->unprocessableHtmlTags ) ) { | |
| 1139 | - return $html; | |
| 1140 | - } | |
| 1141 | - | |
| 1142 | - $unprocessableHtmlTags = implode( '|', $this->unprocessableHtmlTags ); | |
| 1143 | - | |
| 1144 | - return preg_replace( | |
| 1145 | - '/<\\/?(' . $unprocessableHtmlTags . ')[^>]*>/i', | |
| 1146 | - '', | |
| 1147 | - $html | |
| 1148 | - ); | |
| 1149 | - } | |
| 1150 | - | |
| 1151 | - /** | |
| 1152 | - * Makes sure that the passed HTML has a document type. | |
| 1153 | - * | |
| 1154 | - * @param string $html | |
| 1155 | - * | |
| 1156 | - * @return string HTML with document type | |
| 1157 | - */ | |
| 1158 | - private function ensureDocumentType( $html ) { | |
| 1159 | - $hasDocumentType = stripos( $html, '<!DOCTYPE' ) !== false; | |
| 1160 | - | |
| 1161 | - if ( $hasDocumentType ) { | |
| 1162 | - return $html; | |
| 1163 | - } | |
| 1164 | - | |
| 1165 | - return self::DEFAULT_DOCUMENT_TYPE . $html; | |
| 1166 | - } | |
| 1167 | - | |
| 1168 | - /** | |
| 1169 | - * Adds a Content-Type meta tag for the charset. | |
| 1170 | - * | |
| 1171 | - * @param string $html | |
| 1172 | - * | |
| 1173 | - * @return string the HTML with the meta tag added | |
| 1174 | - */ | |
| 1175 | - private function addContentTypeMetaTag( $html ) { | |
| 1176 | - $hasContentTypeMetaTag = stristr( $html, 'Content-Type' ) !== false; | |
| 1177 | - | |
| 1178 | - if ( $hasContentTypeMetaTag ) { | |
| 1179 | - return $html; | |
| 1180 | - } | |
| 1181 | - | |
| 1182 | - // We are trying to insert the meta tag to the right spot in the DOM. | |
| 1183 | - // If we just prepended it to the HTML, we would lose attributes set to the HTML tag. | |
| 1184 | - $hasHeadTag = stripos( $html, '<head' ) !== false; | |
| 1185 | - $hasHtmlTag = stripos( $html, '<html' ) !== false; | |
| 1186 | - | |
| 1187 | - if ( $hasHeadTag ) { | |
| 1188 | - $reworkedHtml = preg_replace( '/<head(.*?)>/i', '<head$1>' . self::CONTENT_TYPE_META_TAG, $html ); | |
| 1189 | - } elseif ( $hasHtmlTag ) { | |
| 1190 | - $reworkedHtml = preg_replace( | |
| 1191 | - '/<html(.*?)>/i', | |
| 1192 | - '<html$1><head>' . self::CONTENT_TYPE_META_TAG . '</head>', | |
| 1193 | - $html | |
| 1194 | - ); | |
| 1195 | - } else { | |
| 1196 | - $reworkedHtml = self::CONTENT_TYPE_META_TAG . $html; | |
| 1197 | - } | |
| 1198 | - | |
| 1199 | - return $reworkedHtml; | |
| 1200 | - } | |
| 1201 | - | |
| 1202 | - /** | |
| 1203 | - * @param string[] $a | |
| 1204 | - * @param string[] $b | |
| 1205 | - * | |
| 1206 | - * @return int | |
| 1207 | - */ | |
| 1208 | - private function sortBySelectorPrecedence( array $a, array $b ) { | |
| 1209 | - $precedenceA = $this->getCssSelectorPrecedence( $a['selector'] ); | |
| 1210 | - $precedenceB = $this->getCssSelectorPrecedence( $b['selector'] ); | |
| 1211 | - | |
| 1212 | - // We want these sorted in ascending order so selectors with lesser precedence get processed first and | |
| 1213 | - // selectors with greater precedence get sorted last. | |
| 1214 | - $precedenceForEquals = ( $a['line'] < $b['line'] ? -1 : 1 ); | |
| 1215 | - $precedenceForNotEquals = ( $precedenceA < $precedenceB ? -1 : 1 ); | |
| 1216 | - | |
| 1217 | - return ( $precedenceA === $precedenceB ) ? $precedenceForEquals : $precedenceForNotEquals; | |
| 1218 | - } | |
| 1219 | - | |
| 1220 | - /** | |
| 1221 | - * @param string $selector | |
| 1222 | - * | |
| 1223 | - * @return int | |
| 1224 | - */ | |
| 1225 | - private function getCssSelectorPrecedence( $selector ) { | |
| 1226 | - $selectorKey = md5( $selector ); | |
| 1227 | - | |
| 1228 | - if ( !isset( $this->caches[self::CACHE_KEY_SELECTOR][$selectorKey] ) ) { | |
| 1229 | - $precedence = 0; | |
| 1230 | - $value = 100; | |
| 1231 | - // ids: worth 100, classes: worth 10, elements: worth 1 | |
| 1232 | - $search = ['\\#', '\\.', '']; | |
| 1233 | - | |
| 1234 | - foreach ( $search as $s ) { | |
| 1235 | - if ( trim( $selector ) === '' ) { | |
| 1236 | - break; | |
| 1237 | - } | |
| 1238 | - $number = 0; | |
| 1239 | - $selector = preg_replace( '/' . $s . '\\w+/', '', $selector, -1, $number ); | |
| 1240 | - $precedence += ( $value * $number ); | |
| 1241 | - $value /= 10; | |
| 1242 | - } | |
| 1243 | - $this->caches[self::CACHE_KEY_SELECTOR][$selectorKey] = $precedence; | |
| 1244 | - } | |
| 1245 | - | |
| 1246 | - return $this->caches[self::CACHE_KEY_SELECTOR][$selectorKey]; | |
| 1247 | - } | |
| 1248 | - | |
| 1249 | - /** | |
| 1250 | - * Maps a CSS selector to an XPath query string. | |
| 1251 | - * | |
| 1252 | - * @see http://plasmasturm.org/log/444/ | |
| 1253 | - * | |
| 1254 | - * @param string $cssSelector a CSS selector | |
| 1255 | - * | |
| 1256 | - * @return string the corresponding XPath selector | |
| 1257 | - */ | |
| 1258 | - private function translateCssToXpath( $cssSelector ) { | |
| 1259 | - $paddedSelector = ' ' . $cssSelector . ' '; | |
| 1260 | - $lowercasePaddedSelector = preg_replace_callback( | |
| 1261 | - '/\\s+\\w+\\s+/', | |
| 1262 | - function ( array $matches ) { | |
| 1263 | - return strtolower( $matches[0] ); | |
| 1264 | - }, | |
| 1265 | - $paddedSelector | |
| 1266 | - ); | |
| 1267 | - $trimmedLowercaseSelector = trim( $lowercasePaddedSelector ); | |
| 1268 | - $xPathKey = md5( $trimmedLowercaseSelector ); | |
| 1269 | - | |
| 1270 | - if ( !isset( $this->caches[self::CACHE_KEY_XPATH][$xPathKey] ) ) { | |
| 1271 | - $roughXpath = '//' . preg_replace( | |
| 1272 | - array_keys( $this->xPathRules ), | |
| 1273 | - $this->xPathRules, | |
| 1274 | - $trimmedLowercaseSelector | |
| 1275 | - ); | |
| 1276 | - $xPathWithIdAttributeMatchers = preg_replace_callback( | |
| 1277 | - self::ID_ATTRIBUTE_MATCHER, | |
| 1278 | - [$this, 'matchIdAttributes'], | |
| 1279 | - $roughXpath | |
| 1280 | - ); | |
| 1281 | - $xPathWithIdAttributeAndClassMatchers = preg_replace_callback( | |
| 1282 | - self::CLASS_ATTRIBUTE_MATCHER, | |
| 1283 | - [$this, 'matchClassAttributes'], | |
| 1284 | - $xPathWithIdAttributeMatchers | |
| 1285 | - ); | |
| 1286 | - | |
| 1287 | - // Advanced selectors are going to require a bit more advanced emogrification. | |
| 1288 | - // When we required PHP 5.3, we could do this with closures. | |
| 1289 | - $xPathWithIdAttributeAndClassMatchers = preg_replace_callback( | |
| 1290 | - '/([^\\/]+):nth-child\\(\\s*(odd|even|[+\\-]?\\d|[+\\-]?\\d?n(\\s*[+\\-]\\s*\\d)?)\\s*\\)/i', | |
| 1291 | - [$this, 'translateNthChild'], | |
| 1292 | - $xPathWithIdAttributeAndClassMatchers | |
| 1293 | - ); | |
| 1294 | - $finalXpath = preg_replace_callback( | |
| 1295 | - '/([^\\/]+):nth-of-type\\(\s*(odd|even|[+\\-]?\\d|[+\\-]?\\d?n(\\s*[+\\-]\\s*\\d)?)\\s*\\)/i', | |
| 1296 | - [$this, 'translateNthOfType'], | |
| 1297 | - $xPathWithIdAttributeAndClassMatchers | |
| 1298 | - ); | |
| 1299 | - | |
| 1300 | - $this->caches[self::CACHE_KEY_SELECTOR][$xPathKey] = $finalXpath; | |
| 1301 | - } | |
| 1302 | - | |
| 1303 | - return $this->caches[self::CACHE_KEY_SELECTOR][$xPathKey]; | |
| 1304 | - } | |
| 1305 | - | |
| 1306 | - /** | |
| 1307 | - * @param string[] $match | |
| 1308 | - * | |
| 1309 | - * @return string | |
| 1310 | - */ | |
| 1311 | - private function matchIdAttributes( array $match ) { | |
| 1312 | - return ( $match[1] !== '' ? $match[1] : '*' ) . '[@id="' . $match[2] . '"]'; | |
| 1313 | - } | |
| 1314 | - | |
| 1315 | - /** | |
| 1316 | - * @param string[] $match | |
| 1317 | - * | |
| 1318 | - * @return string | |
| 1319 | - */ | |
| 1320 | - private function matchClassAttributes( array $match ) { | |
| 1321 | - return ( $match[1] !== '' ? $match[1] : '*' ) . '[contains(concat(" ",@class," "),concat(" ","' . | |
| 1322 | - implode( | |
| 1323 | - '"," "))][contains(concat(" ",@class," "),concat(" ","', | |
| 1324 | - explode( '.', substr( $match[2], 1 ) ) | |
| 1325 | - ) . '"," "))]'; | |
| 1326 | - } | |
| 1327 | - | |
| 1328 | - /** | |
| 1329 | - * @param string[] $match | |
| 1330 | - * | |
| 1331 | - * @return string | |
| 1332 | - */ | |
| 1333 | - private function translateNthChild( array $match ) { | |
| 1334 | - $parseResult = $this->parseNth( $match ); | |
| 1335 | - | |
| 1336 | - if ( isset( $parseResult[self::MULTIPLIER] ) ) { | |
| 1337 | - if ( $parseResult[self::MULTIPLIER] < 0 ) { | |
| 1338 | - $parseResult[self::MULTIPLIER] = abs( $parseResult[self::MULTIPLIER] ); | |
| 1339 | - $xPathExpression = sprintf( | |
| 1340 | - '*[(last() - position()) mod %u = %u]/self::%s', | |
| 1341 | - $parseResult[self::MULTIPLIER], | |
| 1342 | - $parseResult[self::INDEX], | |
| 1343 | - $match[1] | |
| 1344 | - ); | |
| 1345 | - } else { | |
| 1346 | - $xPathExpression = sprintf( | |
| 1347 | - '*[position() mod %u = %u]/self::%s', | |
| 1348 | - $parseResult[self::MULTIPLIER], | |
| 1349 | - $parseResult[self::INDEX], | |
| 1350 | - $match[1] | |
| 1351 | - ); | |
| 1352 | - } | |
| 1353 | - } else { | |
| 1354 | - $xPathExpression = sprintf( '*[%u]/self::%s', $parseResult[self::INDEX], $match[1] ); | |
| 1355 | - } | |
| 1356 | - | |
| 1357 | - return $xPathExpression; | |
| 1358 | - } | |
| 1359 | - | |
| 1360 | - /** | |
| 1361 | - * @param string[] $match | |
| 1362 | - * | |
| 1363 | - * @return string | |
| 1364 | - */ | |
| 1365 | - private function translateNthOfType( array $match ) { | |
| 1366 | - $parseResult = $this->parseNth( $match ); | |
| 1367 | - | |
| 1368 | - if ( isset( $parseResult[self::MULTIPLIER] ) ) { | |
| 1369 | - if ( $parseResult[self::MULTIPLIER] < 0 ) { | |
| 1370 | - $parseResult[self::MULTIPLIER] = abs( $parseResult[self::MULTIPLIER] ); | |
| 1371 | - $xPathExpression = sprintf( | |
| 1372 | - '%s[(last() - position()) mod %u = %u]', | |
| 1373 | - $match[1], | |
| 1374 | - $parseResult[self::MULTIPLIER], | |
| 1375 | - $parseResult[self::INDEX] | |
| 1376 | - ); | |
| 1377 | - } else { | |
| 1378 | - $xPathExpression = sprintf( | |
| 1379 | - '%s[position() mod %u = %u]', | |
| 1380 | - $match[1], | |
| 1381 | - $parseResult[self::MULTIPLIER], | |
| 1382 | - $parseResult[self::INDEX] | |
| 1383 | - ); | |
| 1384 | - } | |
| 1385 | - } else { | |
| 1386 | - $xPathExpression = sprintf( '%s[%u]', $match[1], $parseResult[self::INDEX] ); | |
| 1387 | - } | |
| 1388 | - | |
| 1389 | - return $xPathExpression; | |
| 1390 | - } | |
| 1391 | - | |
| 1392 | - /** | |
| 1393 | - * @param string[] $match | |
| 1394 | - * | |
| 1395 | - * @return int[] | |
| 1396 | - */ | |
| 1397 | - private function parseNth( array $match ) { | |
| 1398 | - if ( in_array( strtolower( $match[2] ), ['even', 'odd'], true ) ) { | |
| 1399 | - // we have "even" or "odd" | |
| 1400 | - $index = strtolower( $match[2] ) === 'even' ? 0 : 1; | |
| 1401 | - | |
| 1402 | - return [self::MULTIPLIER => 2, self::INDEX => $index]; | |
| 1403 | - } | |
| 1404 | - | |
| 1405 | - if ( stripos( $match[2], 'n' ) === false ) { | |
| 1406 | - // if there is a multiplier | |
| 1407 | - $index = (int) str_replace( ' ', '', $match[2] ); | |
| 1408 | - | |
| 1409 | - return [self::INDEX => $index]; | |
| 1410 | - } | |
| 1411 | - | |
| 1412 | - if ( isset( $match[3] ) ) { | |
| 1413 | - $multipleTerm = str_replace( $match[3], '', $match[2] ); | |
| 1414 | - $index = (int) str_replace( ' ', '', $match[3] ); | |
| 1415 | - } else { | |
| 1416 | - $multipleTerm = $match[2]; | |
| 1417 | - $index = 0; | |
| 1418 | - } | |
| 1419 | - | |
| 1420 | - $multiplier = str_ireplace( 'n', '', $multipleTerm ); | |
| 1421 | - | |
| 1422 | - if ( $multiplier === '' ) { | |
| 1423 | - $multiplier = 1; | |
| 1424 | - } elseif ( $multiplier === '0' ) { | |
| 1425 | - return [self::INDEX => $index]; | |
| 1426 | - } else { | |
| 1427 | - $multiplier = (int) $multiplier; | |
| 1428 | - } | |
| 1429 | - | |
| 1430 | - while ( $index < 0 ) { | |
| 1431 | - $index += abs( $multiplier ); | |
| 1432 | - } | |
| 1433 | - | |
| 1434 | - return [self::MULTIPLIER => $multiplier, self::INDEX => $index]; | |
| 1435 | - } | |
| 1436 | - | |
| 1437 | - /** | |
| 1438 | - * Parses a CSS declaration block into property name/value pairs. | |
| 1439 | - * | |
| 1440 | - * Example: | |
| 1441 | - * | |
| 1442 | - * The declaration block | |
| 1443 | - * | |
| 1444 | - * "color: #000; font-weight: bold;" | |
| 1445 | - * | |
| 1446 | - * will be parsed into the following array: | |
| 1447 | - * | |
| 1448 | - * "color" => "#000" | |
| 1449 | - * "font-weight" => "bold" | |
| 1450 | - * | |
| 1451 | - * @param string $cssDeclarationsBlock the CSS declarations block without the curly braces, may be empty | |
| 1452 | - * | |
| 1453 | - * @return string[] | |
| 1454 | - * the CSS declarations with the property names as array keys and the property values as array values | |
| 1455 | - */ | |
| 1456 | - private function parseCssDeclarationsBlock( $cssDeclarationsBlock ) { | |
| 1457 | - if ( isset( $this->caches[self::CACHE_KEY_CSS_DECLARATIONS_BLOCK][$cssDeclarationsBlock] ) ) { | |
| 1458 | - return $this->caches[self::CACHE_KEY_CSS_DECLARATIONS_BLOCK][$cssDeclarationsBlock]; | |
| 1459 | - } | |
| 1460 | - | |
| 1461 | - $properties = []; | |
| 1462 | - $declarations = preg_split( '/;(?!base64|charset)/', $cssDeclarationsBlock ); | |
| 1463 | - | |
| 1464 | - foreach ( $declarations as $declaration ) { | |
| 1465 | - $matches = []; | |
| 1466 | - | |
| 1467 | - if ( !preg_match( '/^([A-Za-z\\-]+)\\s*:\\s*(.+)$/', trim( $declaration ), $matches ) ) { | |
| 1468 | - continue; | |
| 1469 | - } | |
| 1470 | - | |
| 1471 | - $propertyName = strtolower( $matches[1] ); | |
| 1472 | - $propertyValue = $matches[2]; | |
| 1473 | - $properties[$propertyName] = $propertyValue; | |
| 1474 | - } | |
| 1475 | - $this->caches[self::CACHE_KEY_CSS_DECLARATIONS_BLOCK][$cssDeclarationsBlock] = $properties; | |
| 1476 | - | |
| 1477 | - return $properties; | |
| 1478 | - } | |
| 1479 | - | |
| 1480 | - /** | |
| 1481 | - * Find the nodes that are not to be emogrified. | |
| 1482 | - * | |
| 1483 | - * @param \DOMXPath $xPath | |
| 1484 | - * | |
| 1485 | - * @return \DOMElement[] | |
| 1486 | - */ | |
| 1487 | - private function getNodesToExclude( DOMXPath $xPath ) { | |
| 1488 | - $excludedNodes = []; | |
| 1489 | - | |
| 1490 | - foreach ( array_keys( $this->excludedSelectors ) as $selectorToExclude ) { | |
| 1491 | - foreach ( $xPath->query( $this->translateCssToXpath( $selectorToExclude ) ) as $node ) { | |
| 1492 | - $excludedNodes[] = $node; | |
| 1493 | - } | |
| 1494 | - } | |
| 1495 | - | |
| 1496 | - return $excludedNodes; | |
| 1497 | - } | |
| 1498 | - | |
| 1499 | - /** | |
| 1500 | - * Handles invalid xPath expression warnings, generated by process() method, | |
| 1501 | - * during querying \DOMDocument and trigger \InvalidArgumentException | |
| 1502 | - * with invalid selector. | |
| 1503 | - * | |
| 1504 | - * @param int $type | |
| 1505 | - * @param string $message | |
| 1506 | - * @param string $file | |
| 1507 | - * @param int $line | |
| 1508 | - * | |
| 1509 | - * @return bool always false | |
| 1510 | - * | |
| 1511 | - * @throws InvalidArgumentException | |
| 1512 | - */ | |
| 1513 | - public function handleXpathError( $type, $message, $file, $line, array $context ) { | |
| 1514 | - if ( $this->debug && $type === E_WARNING && isset( $context['cssRule']['selector'] ) ) { | |
| 1515 | - throw new InvalidArgumentException( sprintf( '%s in selector >> %s << in %s on line %s', $message, $context['cssRule']['selector'], $file, $line ) ); | |
| 1516 | - } | |
| 1517 | - | |
| 1518 | - // the normal error handling continues when handler return false | |
| 1519 | - return false; | |
| 1520 | - } | |
| 1521 | - | |
| 1522 | - /** | |
| 1523 | - * Sets the debug mode. | |
| 1524 | - * | |
| 1525 | - * @param bool $debug set to true to enable debug mode | |
| 1526 | - * | |
| 1527 | - * @return void | |
| 1528 | - */ | |
| 1529 | - public function setDebug( $debug ) { | |
| 1530 | - $this->debug = $debug; | |
| 1531 | - } | |
| 1532 | -} | |
| 1 | +<?php | |
| 2 | + | |
| 3 | +/** | |
| 4 | + * This class provides functions for converting CSS styles into inline style attributes in your HTML code. | |
| 5 | + * | |
| 6 | + * For more information, please see the README.md file. | |
| 7 | + * | |
| 8 | + * @version 1.2.0 | |
| 9 | + * | |
| 10 | + * @author Cameron Brooks | |
| 11 | + * @author Jaime Prado | |
| 12 | + * @author Oliver Klee <typo3-coding@oliverklee.de> | |
| 13 | + * @author Roman Ožana <ozana@omdesign.cz> | |
| 14 | + * @author Sander Kruger <s.kruger@invessel.com> | |
| 15 | + */ | |
| 16 | +class Emogrifier | |
| 17 | +{ | |
| 18 | + /** | |
| 19 | + * @var int | |
| 20 | + */ | |
| 21 | + const CACHE_KEY_CSS = 0; | |
| 22 | + | |
| 23 | + /** | |
| 24 | + * @var int | |
| 25 | + */ | |
| 26 | + const CACHE_KEY_SELECTOR = 1; | |
| 27 | + | |
| 28 | + /** | |
| 29 | + * @var int | |
| 30 | + */ | |
| 31 | + const CACHE_KEY_XPATH = 2; | |
| 32 | + | |
| 33 | + /** | |
| 34 | + * @var int | |
| 35 | + */ | |
| 36 | + const CACHE_KEY_CSS_DECLARATIONS_BLOCK = 3; | |
| 37 | + | |
| 38 | + /** | |
| 39 | + * @var int | |
| 40 | + */ | |
| 41 | + const CACHE_KEY_COMBINED_STYLES = 4; | |
| 42 | + | |
| 43 | + /** | |
| 44 | + * for calculating nth-of-type and nth-child selectors | |
| 45 | + * | |
| 46 | + * @var int | |
| 47 | + */ | |
| 48 | + const INDEX = 0; | |
| 49 | + | |
| 50 | + /** | |
| 51 | + * for calculating nth-of-type and nth-child selectors | |
| 52 | + * | |
| 53 | + * @var int | |
| 54 | + */ | |
| 55 | + const MULTIPLIER = 1; | |
| 56 | + | |
| 57 | + /** | |
| 58 | + * @var string | |
| 59 | + */ | |
| 60 | + const ID_ATTRIBUTE_MATCHER = '/(\\w+)?\\#([\\w\\-]+)/'; | |
| 61 | + | |
| 62 | + /** | |
| 63 | + * @var string | |
| 64 | + */ | |
| 65 | + const CLASS_ATTRIBUTE_MATCHER = '/(\\w+|[\\*\\]])?((\\.[\\w\\-]+)+)/'; | |
| 66 | + | |
| 67 | + /** | |
| 68 | + * @var string | |
| 69 | + */ | |
| 70 | + const CONTENT_TYPE_META_TAG = '<meta http-equiv="Content-Type" content="text/html; charset=utf-8">'; | |
| 71 | + | |
| 72 | + /** | |
| 73 | + * @var string | |
| 74 | + */ | |
| 75 | + const DEFAULT_DOCUMENT_TYPE = '<!DOCTYPE html>'; | |
| 76 | + | |
| 77 | + /** | |
| 78 | + * @var string | |
| 79 | + */ | |
| 80 | + private $html = ''; | |
| 81 | + | |
| 82 | + /** | |
| 83 | + * @var string | |
| 84 | + */ | |
| 85 | + private $css = ''; | |
| 86 | + | |
| 87 | + /** | |
| 88 | + * @var bool[] | |
| 89 | + */ | |
| 90 | + private $excludedSelectors = []; | |
| 91 | + | |
| 92 | + /** | |
| 93 | + * @var string[] | |
| 94 | + */ | |
| 95 | + private $unprocessableHtmlTags = ['wbr']; | |
| 96 | + | |
| 97 | + /** | |
| 98 | + * @var bool[] | |
| 99 | + */ | |
| 100 | + private $allowedMediaTypes = ['all' => true, 'screen' => true, 'print' => true]; | |
| 101 | + | |
| 102 | + /** | |
| 103 | + * @var mixed[] | |
| 104 | + */ | |
| 105 | + private $caches = [ | |
| 106 | + self::CACHE_KEY_CSS => [], | |
| 107 | + self::CACHE_KEY_SELECTOR => [], | |
| 108 | + self::CACHE_KEY_XPATH => [], | |
| 109 | + self::CACHE_KEY_CSS_DECLARATIONS_BLOCK => [], | |
| 110 | + self::CACHE_KEY_COMBINED_STYLES => [], | |
| 111 | + ]; | |
| 112 | + | |
| 113 | + /** | |
| 114 | + * the visited nodes with the XPath paths as array keys | |
| 115 | + * | |
| 116 | + * @var \DOMElement[] | |
| 117 | + */ | |
| 118 | + private $visitedNodes = []; | |
| 119 | + | |
| 120 | + /** | |
| 121 | + * the styles to apply to the nodes with the XPath paths as array keys for the outer array | |
| 122 | + * and the attribute names/values as key/value pairs for the inner array | |
| 123 | + * | |
| 124 | + * @var string[][] | |
| 125 | + */ | |
| 126 | + private $styleAttributesForNodes = []; | |
| 127 | + | |
| 128 | + /** | |
| 129 | + * Determines whether the "style" attributes of tags in the the HTML passed to this class should be preserved. | |
| 130 | + * If set to false, the value of the style attributes will be discarded. | |
| 131 | + * | |
| 132 | + * @var bool | |
| 133 | + */ | |
| 134 | + private $isInlineStyleAttributesParsingEnabled = true; | |
| 135 | + | |
| 136 | + /** | |
| 137 | + * Determines whether the <style> blocks in the HTML passed to this class should be parsed. | |
| 138 | + * | |
| 139 | + * If set to true, the <style> blocks will be removed from the HTML and their contents will be applied to the HTML | |
| 140 | + * via inline styles. | |
| 141 | + * | |
| 142 | + * If set to false, the <style> blocks will be left as they are in the HTML. | |
| 143 | + * | |
| 144 | + * @var bool | |
| 145 | + */ | |
| 146 | + private $isStyleBlocksParsingEnabled = true; | |
| 147 | + | |
| 148 | + /** | |
| 149 | + * Determines whether elements with the `display: none` property are | |
| 150 | + * removed from the DOM. | |
| 151 | + * | |
| 152 | + * @var bool | |
| 153 | + */ | |
| 154 | + private $shouldKeepInvisibleNodes = true; | |
| 155 | + | |
| 156 | + /** | |
| 157 | + * @var string[] | |
| 158 | + */ | |
| 159 | + private $xPathRules = [ | |
| 160 | + // child | |
| 161 | + '/\\s*>\\s*/' => '/', | |
| 162 | + // adjacent sibling | |
| 163 | + '/\\s+\\+\\s+/' => '/following-sibling::*[1]/self::', | |
| 164 | + // descendant | |
| 165 | + '/\\s+(?=.*[^\\]]{1}$)/' => '//', | |
| 166 | + // :first-child | |
| 167 | + '/([^\\/]+):first-child/i' => '*[1]/self::\\1', | |
| 168 | + // :last-child | |
| 169 | + '/([^\\/]+):last-child/i' => '*[last()]/self::\\1', | |
| 170 | + // attribute only | |
| 171 | + '/^\\[(\\w+|\\w+\\=[\'"]?\\w+[\'"]?)\\]/' => '*[@\\1]', | |
| 172 | + // attribute | |
| 173 | + '/(\\w)\\[(\\w+)\\]/' => '\\1[@\\2]', | |
| 174 | + // exact attribute | |
| 175 | + '/(\\w)\\[(\\w+)\\=[\'"]?([\\w\\s]+)[\'"]?\\]/' => '\\1[@\\2="\\3"]', | |
| 176 | + // element attribute~= | |
| 177 | + '/([\\w\\*]+)\\[(\\w+)[\\s]*\\~\\=[\\s]*[\'"]?([\\w-_\\/]+)[\'"]?\\]/' | |
| 178 | + => '\\1[contains(concat(" ", @\\2, " "), concat(" ", "\\3", " "))]', | |
| 179 | + // element attribute^= | |
| 180 | + '/([\\w\\*]+)\\[(\\w+)[\\s]*\\^\\=[\\s]*[\'"]?([\\w-_\\/]+)[\'"]?\\]/' => '\\1[starts-with(@\\2, "\\3")]', | |
| 181 | + // element attribute*= | |
| 182 | + '/([\\w\\*]+)\\[(\\w+)[\\s]*\\*\\=[\\s]*[\'"]?([\\w-_\\s\\/:;]+)[\'"]?\\]/' => '\\1[contains(@\\2, "\\3")]', | |
| 183 | + // element attribute$= | |
| 184 | + '/([\\w\\*]+)\\[(\\w+)[\\s]*\\$\\=[\\s]*[\'"]?([\\w-_\\s\\/]+)[\'"]?\\]/' | |
| 185 | + => '\\1[substring(@\\2, string-length(@\\2) - string-length("\\3") + 1) = "\\3"]', | |
| 186 | + // element attribute|= | |
| 187 | + '/([\\w\\*]+)\\[(\\w+)[\\s]*\\|\\=[\\s]*[\'"]?([\\w-_\\s\\/]+)[\'"]?\\]/' | |
| 188 | + => '\\1[@\\2="\\3" or starts-with(@\\2, concat("\\3", "-"))]', | |
| 189 | + ]; | |
| 190 | + | |
| 191 | + /** | |
| 192 | + * Determines whether CSS styles that have an equivalent HTML attribute | |
| 193 | + * should be mapped and attached to those elements. | |
| 194 | + * | |
| 195 | + * @var bool | |
| 196 | + */ | |
| 197 | + private $shouldMapCssToHtml = false; | |
| 198 | + | |
| 199 | + /** | |
| 200 | + * This multi-level array contains simple mappings of CSS properties to | |
| 201 | + * HTML attributes. If a mapping only applies to certain HTML nodes or | |
| 202 | + * only for certain values, the mapping is an object with a whitelist | |
| 203 | + * of nodes and values. | |
| 204 | + * | |
| 205 | + * @var mixed[][] | |
| 206 | + */ | |
| 207 | + private $cssToHtmlMap = [ | |
| 208 | + 'background-color' => [ | |
| 209 | + 'attribute' => 'bgcolor', | |
| 210 | + ], | |
| 211 | + 'text-align' => [ | |
| 212 | + 'attribute' => 'align', | |
| 213 | + 'nodes' => ['p', 'div', 'td'], | |
| 214 | + 'values' => ['left', 'right', 'center', 'justify'], | |
| 215 | + ], | |
| 216 | + 'float' => [ | |
| 217 | + 'attribute' => 'align', | |
| 218 | + 'nodes' => ['table', 'img'], | |
| 219 | + 'values' => ['left', 'right'], | |
| 220 | + ], | |
| 221 | + 'border-spacing' => [ | |
| 222 | + 'attribute' => 'cellspacing', | |
| 223 | + 'nodes' => ['table'], | |
| 224 | + ], | |
| 225 | + ]; | |
| 226 | + | |
| 227 | + /** | |
| 228 | + * Emogrifier will throw Exceptions when it encounters an error instead of silently ignoring them. | |
| 229 | + * | |
| 230 | + * @var bool | |
| 231 | + */ | |
| 232 | + private $debug = false; | |
| 233 | + | |
| 234 | + /** | |
| 235 | + * The constructor. | |
| 236 | + * | |
| 237 | + * @param string $html the HTML to emogrify, must be UTF-8-encoded | |
| 238 | + * @param string $css the CSS to merge, must be UTF-8-encoded | |
| 239 | + */ | |
| 240 | + public function __construct($html = '', $css = '') | |
| 241 | + { | |
| 242 | + $this->setHtml($html); | |
| 243 | + $this->setCss($css); | |
| 244 | + } | |
| 245 | + | |
| 246 | + /** | |
| 247 | + * The destructor. | |
| 248 | + */ | |
| 249 | + public function __destruct() | |
| 250 | + { | |
| 251 | + $this->purgeVisitedNodes(); | |
| 252 | + } | |
| 253 | + | |
| 254 | + /** | |
| 255 | + * Sets the HTML to emogrify. | |
| 256 | + * | |
| 257 | + * @param string $html the HTML to emogrify, must be UTF-8-encoded | |
| 258 | + * | |
| 259 | + * @return void | |
| 260 | + */ | |
| 261 | + public function setHtml($html) | |
| 262 | + { | |
| 263 | + $this->html = $html; | |
| 264 | + } | |
| 265 | + | |
| 266 | + /** | |
| 267 | + * Sets the CSS to merge with the HTML. | |
| 268 | + * | |
| 269 | + * @param string $css the CSS to merge, must be UTF-8-encoded | |
| 270 | + * | |
| 271 | + * @return void | |
| 272 | + */ | |
| 273 | + public function setCss($css) | |
| 274 | + { | |
| 275 | + $this->css = $css; | |
| 276 | + } | |
| 277 | + | |
| 278 | + /** | |
| 279 | + * Applies $this->css to $this->html and returns the HTML with the CSS | |
| 280 | + * applied. | |
| 281 | + * | |
| 282 | + * This method places the CSS inline. | |
| 283 | + * | |
| 284 | + * @return string | |
| 285 | + * | |
| 286 | + * @throws \BadMethodCallException | |
| 287 | + */ | |
| 288 | + public function emogrify() | |
| 289 | + { | |
| 290 | + if ($this->html === '') { | |
| 291 | + throw new \BadMethodCallException('Please set some HTML first before calling emogrify.', 1390393096); | |
| 292 | + } | |
| 293 | + | |
| 294 | + $xmlDocument = $this->createXmlDocument(); | |
| 295 | + $this->process($xmlDocument); | |
| 296 | + | |
| 297 | + return $xmlDocument->saveHTML(); | |
| 298 | + } | |
| 299 | + | |
| 300 | + /** | |
| 301 | + * Applies $this->css to $this->html and returns only the HTML content | |
| 302 | + * within the <body> tag. | |
| 303 | + * | |
| 304 | + * This method places the CSS inline. | |
| 305 | + * | |
| 306 | + * @return string | |
| 307 | + * | |
| 308 | + * @throws \BadMethodCallException | |
| 309 | + */ | |
| 310 | + public function emogrifyBodyContent() | |
| 311 | + { | |
| 312 | + if ($this->html === '') { | |
| 313 | + throw new \BadMethodCallException('Please set some HTML first before calling emogrify.', 1390393096); | |
| 314 | + } | |
| 315 | + | |
| 316 | + $xmlDocument = $this->createXmlDocument(); | |
| 317 | + $this->process($xmlDocument); | |
| 318 | + | |
| 319 | + $innerDocument = new \DOMDocument(); | |
| 320 | + foreach ($xmlDocument->documentElement->getElementsByTagName('body')->item(0)->childNodes as $childNode) { | |
| 321 | + $innerDocument->appendChild($innerDocument->importNode($childNode, true)); | |
| 322 | + } | |
| 323 | + | |
| 324 | + return html_entity_decode($innerDocument->saveHTML()); | |
| 325 | + } | |
| 326 | + | |
| 327 | + /** | |
| 328 | + * Applies $this->css to $xmlDocument. | |
| 329 | + * | |
| 330 | + * This method places the CSS inline. | |
| 331 | + * | |
| 332 | + * @param \DOMDocument $xmlDocument | |
| 333 | + * | |
| 334 | + * @return void | |
| 335 | + * | |
| 336 | + * @throws \InvalidArgumentException | |
| 337 | + */ | |
| 338 | + protected function process(\DOMDocument $xmlDocument) | |
| 339 | + { | |
| 340 | + $xPath = new \DOMXPath($xmlDocument); | |
| 341 | + $this->clearAllCaches(); | |
| 342 | + | |
| 343 | + // Before be begin processing the CSS file, parse the document and normalize all existing CSS attributes. | |
| 344 | + // This changes 'DISPLAY: none' to 'display: none'. | |
| 345 | + // We wouldn't have to do this if DOMXPath supported XPath 2.0. | |
| 346 | + // Also store a reference of nodes with existing inline styles so we don't overwrite them. | |
| 347 | + $this->purgeVisitedNodes(); | |
| 348 | + | |
| 349 | + set_error_handler([$this, 'handleXpathError'], E_WARNING); | |
| 350 | + | |
| 351 | + $nodesWithStyleAttributes = $xPath->query('//*[@style]'); | |
| 352 | + if ($nodesWithStyleAttributes !== false) { | |
| 353 | + /** @var \DOMElement $node */ | |
| 354 | + foreach ($nodesWithStyleAttributes as $node) { | |
| 355 | + if ($this->isInlineStyleAttributesParsingEnabled) { | |
| 356 | + $this->normalizeStyleAttributes($node); | |
| 357 | + } else { | |
| 358 | + $node->removeAttribute('style'); | |
| 359 | + } | |
| 360 | + } | |
| 361 | + } | |
| 362 | + | |
| 363 | + // grab any existing style blocks from the html and append them to the existing CSS | |
| 364 | + // (these blocks should be appended so as to have precedence over conflicting styles in the existing CSS) | |
| 365 | + $allCss = $this->css; | |
| 366 | + | |
| 367 | + if ($this->isStyleBlocksParsingEnabled) { | |
| 368 | + $allCss .= $this->getCssFromAllStyleNodes($xPath); | |
| 369 | + } | |
| 370 | + | |
| 371 | + $cssParts = $this->splitCssAndMediaQuery($allCss); | |
| 372 | + $excludedNodes = $this->getNodesToExclude($xPath); | |
| 373 | + $cssRules = $this->parseCssRules($cssParts['css']); | |
| 374 | + foreach ($cssRules as $cssRule) { | |
| 375 | + // query the body for the xpath selector | |
| 376 | + $nodesMatchingCssSelectors = $xPath->query($this->translateCssToXpath($cssRule['selector'])); | |
| 377 | + // ignore invalid selectors | |
| 378 | + if ($nodesMatchingCssSelectors === false) { | |
| 379 | + continue; | |
| 380 | + } | |
| 381 | + | |
| 382 | + /** @var \DOMElement $node */ | |
| 383 | + foreach ($nodesMatchingCssSelectors as $node) { | |
| 384 | + if (in_array($node, $excludedNodes, true)) { | |
| 385 | + continue; | |
| 386 | + } | |
| 387 | + | |
| 388 | + // if it has a style attribute, get it, process it, and append (overwrite) new stuff | |
| 389 | + if ($node->hasAttribute('style')) { | |
| 390 | + // break it up into an associative array | |
| 391 | + $oldStyleDeclarations = $this->parseCssDeclarationsBlock($node->getAttribute('style')); | |
| 392 | + } else { | |
| 393 | + $oldStyleDeclarations = []; | |
| 394 | + } | |
| 395 | + $newStyleDeclarations = $this->parseCssDeclarationsBlock($cssRule['declarationsBlock']); | |
| 396 | + if ($this->shouldMapCssToHtml) { | |
| 397 | + $this->mapCssToHtmlAttributes($newStyleDeclarations, $node); | |
| 398 | + } | |
| 399 | + $node->setAttribute( | |
| 400 | + 'style', | |
| 401 | + $this->generateStyleStringFromDeclarationsArrays($oldStyleDeclarations, $newStyleDeclarations) | |
| 402 | + ); | |
| 403 | + } | |
| 404 | + } | |
| 405 | + | |
| 406 | + restore_error_handler(); | |
| 407 | + | |
| 408 | + if ($this->isInlineStyleAttributesParsingEnabled) { | |
| 409 | + $this->fillStyleAttributesWithMergedStyles(); | |
| 410 | + } | |
| 411 | + | |
| 412 | + if ($this->shouldKeepInvisibleNodes) { | |
| 413 | + $this->removeInvisibleNodes($xPath); | |
| 414 | + } | |
| 415 | + | |
| 416 | + $this->copyCssWithMediaToStyleNode($xmlDocument, $xPath, $cssParts['media']); | |
| 417 | + } | |
| 418 | + | |
| 419 | + /** | |
| 420 | + * Applies $styles to $node. | |
| 421 | + * | |
| 422 | + * This method maps CSS styles to HTML attributes and adds those to the | |
| 423 | + * node. | |
| 424 | + * | |
| 425 | + * @param string[] $styles the new CSS styles taken from the global styles to be applied to this node | |
| 426 | + * @param \DOMNode $node node to apply styles to | |
| 427 | + * | |
| 428 | + * @return void | |
| 429 | + */ | |
| 430 | + private function mapCssToHtmlAttributes(array $styles, \DOMNode $node) | |
| 431 | + { | |
| 432 | + foreach ($styles as $property => $value) { | |
| 433 | + // Strip !important indicator | |
| 434 | + $value = trim(str_replace('!important', '', $value)); | |
| 435 | + $this->mapCssToHtmlAttribute($property, $value, $node); | |
| 436 | + } | |
| 437 | + } | |
| 438 | + | |
| 439 | + /** | |
| 440 | + * Tries to apply the CSS style to $node as an attribute. | |
| 441 | + * | |
| 442 | + * This method maps a CSS rule to HTML attributes and adds those to the node. | |
| 443 | + * | |
| 444 | + * @param string $property the name of the CSS property to map | |
| 445 | + * @param string $value the value of the style rule to map | |
| 446 | + * @param \DOMNode $node node to apply styles to | |
| 447 | + * | |
| 448 | + * @return void | |
| 449 | + */ | |
| 450 | + private function mapCssToHtmlAttribute($property, $value, \DOMNode $node) | |
| 451 | + { | |
| 452 | + if (!$this->mapSimpleCssProperty($property, $value, $node)) { | |
| 453 | + $this->mapComplexCssProperty($property, $value, $node); | |
| 454 | + } | |
| 455 | + } | |
| 456 | + | |
| 457 | + /** | |
| 458 | + * Looks up the CSS property in the mapping table and maps it if it matches the conditions. | |
| 459 | + * | |
| 460 | + * @param string $property the name of the CSS property to map | |
| 461 | + * @param string $value the value of the style rule to map | |
| 462 | + * @param \DOMNode $node node to apply styles to | |
| 463 | + * | |
| 464 | + * @return bool true if the property cab be mapped using the simple mapping table | |
| 465 | + */ | |
| 466 | + private function mapSimpleCssProperty($property, $value, \DOMNode $node) | |
| 467 | + { | |
| 468 | + if (!isset($this->cssToHtmlMap[$property])) { | |
| 469 | + return false; | |
| 470 | + } | |
| 471 | + | |
| 472 | + $mapping = $this->cssToHtmlMap[$property]; | |
| 473 | + $nodesMatch = !isset($mapping['nodes']) || in_array($node->nodeName, $mapping['nodes'], true); | |
| 474 | + $valuesMatch = !isset($mapping['values']) || in_array($value, $mapping['values'], true); | |
| 475 | + if (!$nodesMatch || !$valuesMatch) { | |
| 476 | + return false; | |
| 477 | + } | |
| 478 | + | |
| 479 | + $node->setAttribute($mapping['attribute'], $value); | |
| 480 | + | |
| 481 | + return true; | |
| 482 | + } | |
| 483 | + | |
| 484 | + /** | |
| 485 | + * Maps CSS properties that need special transformation to an HTML attribute. | |
| 486 | + * | |
| 487 | + * @param string $property the name of the CSS property to map | |
| 488 | + * @param string $value the value of the style rule to map | |
| 489 | + * @param \DOMNode $node node to apply styles to | |
| 490 | + * | |
| 491 | + * @return void | |
| 492 | + */ | |
| 493 | + private function mapComplexCssProperty($property, $value, \DOMNode $node) | |
| 494 | + { | |
| 495 | + $nodeName = $node->nodeName; | |
| 496 | + $isTable = $nodeName === 'table'; | |
| 497 | + $isImage = $nodeName === 'img'; | |
| 498 | + $isTableOrImage = $isTable || $isImage; | |
| 499 | + | |
| 500 | + switch ($property) { | |
| 501 | + case 'background': | |
| 502 | + // Parse out the color, if any | |
| 503 | + $styles = explode(' ', $value); | |
| 504 | + $first = $styles[0]; | |
| 505 | + if (!is_numeric(substr($first, 0, 1)) && substr($first, 0, 3) !== 'url') { | |
| 506 | + // This is not a position or image, assume it's a color | |
| 507 | + $node->setAttribute('bgcolor', $first); | |
| 508 | + } | |
| 509 | + break; | |
| 510 | + case 'width': | |
| 511 | + // intentional fall-through | |
| 512 | + case 'height': | |
| 513 | + // Only parse values in px and %, but not values like "auto". | |
| 514 | + if (preg_match('/^\d+(px|%)$/', $value)) { | |
| 515 | + // Remove 'px'. This regex only conserves numbers and % | |
| 516 | + $number = preg_replace('/[^0-9.%]/', '', $value); | |
| 517 | + $node->setAttribute($property, $number); | |
| 518 | + } | |
| 519 | + break; | |
| 520 | + case 'margin': | |
| 521 | + if ($isTableOrImage) { | |
| 522 | + $margins = $this->parseCssShorthandValue($value); | |
| 523 | + if ($margins['left'] === 'auto' && $margins['right'] === 'auto') { | |
| 524 | + $node->setAttribute('align', 'center'); | |
| 525 | + } | |
| 526 | + } | |
| 527 | + break; | |
| 528 | + case 'border': | |
| 529 | + if ($isTableOrImage) { | |
| 530 | + if ($value === 'none' || $value === '0') { | |
| 531 | + $node->setAttribute('border', '0'); | |
| 532 | + } | |
| 533 | + } | |
| 534 | + break; | |
| 535 | + default: | |
| 536 | + } | |
| 537 | + } | |
| 538 | + | |
| 539 | + /** | |
| 540 | + * Parses a shorthand CSS value and splits it into individual values | |
| 541 | + * | |
| 542 | + * @param string $value a string of CSS value with 1, 2, 3 or 4 sizes | |
| 543 | + * For example: padding: 0 auto; | |
| 544 | + * '0 auto' is split into top: 0, left: auto, bottom: 0, | |
| 545 | + * right: auto. | |
| 546 | + * | |
| 547 | + * @return string[] an array of values for top, right, bottom and left (using these as associative array keys) | |
| 548 | + */ | |
| 549 | + private function parseCssShorthandValue($value) | |
| 550 | + { | |
| 551 | + $values = preg_split('/\\s+/', $value); | |
| 552 | + | |
| 553 | + $css = []; | |
| 554 | + $css['top'] = $values[0]; | |
| 555 | + $css['right'] = (count($values) > 1) ? $values[1] : $css['top']; | |
| 556 | + $css['bottom'] = (count($values) > 2) ? $values[2] : $css['top']; | |
| 557 | + $css['left'] = (count($values) > 3) ? $values[3] : $css['right']; | |
| 558 | + | |
| 559 | + return $css; | |
| 560 | + } | |
| 561 | + | |
| 562 | + /** | |
| 563 | + * Extracts and parses the individual rules from a CSS string. | |
| 564 | + * | |
| 565 | + * @param string $css a string of raw CSS code | |
| 566 | + * | |
| 567 | + * @return string[][] an array of string sub-arrays with the keys | |
| 568 | + * "selector" (the CSS selector(s), e.g., "*" or "h1"), | |
| 569 | + * "declarationsBLock" (the semicolon-separated CSS declarations for that selector(s), | |
| 570 | + * e.g., "color: red; height: 4px;"), | |
| 571 | + * and "line" (the line number e.g. 42) | |
| 572 | + */ | |
| 573 | + private function parseCssRules($css) | |
| 574 | + { | |
| 575 | + $cssKey = md5($css); | |
| 576 | + if (!isset($this->caches[self::CACHE_KEY_CSS][$cssKey])) { | |
| 577 | + // process the CSS file for selectors and definitions | |
| 578 | + preg_match_all('/(?:^|[\\s^{}]*)([^{]+){([^}]*)}/mis', $css, $matches, PREG_SET_ORDER); | |
| 579 | + | |
| 580 | + $cssRules = []; | |
| 581 | + /** @var string[] $cssRule */ | |
| 582 | + foreach ($matches as $key => $cssRule) { | |
| 583 | + $cssDeclaration = trim($cssRule[2]); | |
| 584 | + if ($cssDeclaration === '') { | |
| 585 | + continue; | |
| 586 | + } | |
| 587 | + | |
| 588 | + $selectors = explode(',', $cssRule[1]); | |
| 589 | + foreach ($selectors as $selector) { | |
| 590 | + // don't process pseudo-elements and behavioral (dynamic) pseudo-classes; | |
| 591 | + // only allow structural pseudo-classes | |
| 592 | + $hasPseudoElement = strpos($selector, '::') !== false; | |
| 593 | + $hasAnyPseudoClass = (bool) preg_match('/:[a-zA-Z]/', $selector); | |
| 594 | + $hasSupportedPseudoClass = (bool) preg_match('/:\\S+\\-(child|type\\()/i', $selector); | |
| 595 | + if ($hasPseudoElement || ($hasAnyPseudoClass && !$hasSupportedPseudoClass)) { | |
| 596 | + continue; | |
| 597 | + } | |
| 598 | + | |
| 599 | + $cssRules[] = [ | |
| 600 | + 'selector' => trim($selector), | |
| 601 | + 'declarationsBlock' => $cssDeclaration, | |
| 602 | + // keep track of where it appears in the file, since order is important | |
| 603 | + 'line' => $key, | |
| 604 | + ]; | |
| 605 | + } | |
| 606 | + } | |
| 607 | + | |
| 608 | + usort($cssRules, [$this, 'sortBySelectorPrecedence']); | |
| 609 | + | |
| 610 | + $this->caches[self::CACHE_KEY_CSS][$cssKey] = $cssRules; | |
| 611 | + } | |
| 612 | + | |
| 613 | + return $this->caches[self::CACHE_KEY_CSS][$cssKey]; | |
| 614 | + } | |
| 615 | + | |
| 616 | + /** | |
| 617 | + * Disables the parsing of inline styles. | |
| 618 | + * | |
| 619 | + * @return void | |
| 620 | + */ | |
| 621 | + public function disableInlineStyleAttributesParsing() | |
| 622 | + { | |
| 623 | + $this->isInlineStyleAttributesParsingEnabled = false; | |
| 624 | + } | |
| 625 | + | |
| 626 | + /** | |
| 627 | + * Disables the parsing of <style> blocks. | |
| 628 | + * | |
| 629 | + * @return void | |
| 630 | + */ | |
| 631 | + public function disableStyleBlocksParsing() | |
| 632 | + { | |
| 633 | + $this->isStyleBlocksParsingEnabled = false; | |
| 634 | + } | |
| 635 | + | |
| 636 | + /** | |
| 637 | + * Disables the removal of elements with `display: none` properties. | |
| 638 | + * | |
| 639 | + * @return void | |
| 640 | + */ | |
| 641 | + public function disableInvisibleNodeRemoval() | |
| 642 | + { | |
| 643 | + $this->shouldKeepInvisibleNodes = false; | |
| 644 | + } | |
| 645 | + | |
| 646 | + /** | |
| 647 | + * Enables the attachment/override of HTML attributes for which a | |
| 648 | + * corresponding CSS property has been set. | |
| 649 | + * | |
| 650 | + * @return void | |
| 651 | + */ | |
| 652 | + public function enableCssToHtmlMapping() | |
| 653 | + { | |
| 654 | + $this->shouldMapCssToHtml = true; | |
| 655 | + } | |
| 656 | + | |
| 657 | + /** | |
| 658 | + * Clears all caches. | |
| 659 | + * | |
| 660 | + * @return void | |
| 661 | + */ | |
| 662 | + private function clearAllCaches() | |
| 663 | + { | |
| 664 | + $this->clearCache(self::CACHE_KEY_CSS); | |
| 665 | + $this->clearCache(self::CACHE_KEY_SELECTOR); | |
| 666 | + $this->clearCache(self::CACHE_KEY_XPATH); | |
| 667 | + $this->clearCache(self::CACHE_KEY_CSS_DECLARATIONS_BLOCK); | |
| 668 | + $this->clearCache(self::CACHE_KEY_COMBINED_STYLES); | |
| 669 | + } | |
| 670 | + | |
| 671 | + /** | |
| 672 | + * Clears a single cache by key. | |
| 673 | + * | |
| 674 | + * @param int $key the cache key, must be CACHE_KEY_CSS, CACHE_KEY_SELECTOR, CACHE_KEY_XPATH | |
| 675 | + * or CACHE_KEY_CSS_DECLARATION_BLOCK | |
| 676 | + * | |
| 677 | + * @return void | |
| 678 | + * | |
| 679 | + * @throws \InvalidArgumentException | |
| 680 | + */ | |
| 681 | + private function clearCache($key) | |
| 682 | + { | |
| 683 | + $allowedCacheKeys = [ | |
| 684 | + self::CACHE_KEY_CSS, | |
| 685 | + self::CACHE_KEY_SELECTOR, | |
| 686 | + self::CACHE_KEY_XPATH, | |
| 687 | + self::CACHE_KEY_CSS_DECLARATIONS_BLOCK, | |
| 688 | + self::CACHE_KEY_COMBINED_STYLES, | |
| 689 | + ]; | |
| 690 | + if (!in_array($key, $allowedCacheKeys, true)) { | |
| 691 | + throw new \InvalidArgumentException('Invalid cache key: ' . $key, 1391822035); | |
| 692 | + } | |
| 693 | + | |
| 694 | + $this->caches[$key] = []; | |
| 695 | + } | |
| 696 | + | |
| 697 | + /** | |
| 698 | + * Purges the visited nodes. | |
| 699 | + * | |
| 700 | + * @return void | |
| 701 | + */ | |
| 702 | + private function purgeVisitedNodes() | |
| 703 | + { | |
| 704 | + $this->visitedNodes = []; | |
| 705 | + $this->styleAttributesForNodes = []; | |
| 706 | + } | |
| 707 | + | |
| 708 | + /** | |
| 709 | + * Marks a tag for removal. | |
| 710 | + * | |
| 711 | + * There are some HTML tags that DOMDocument cannot process, and it will throw an error if it encounters them. | |
| 712 | + * In particular, DOMDocument will complain if you try to use HTML5 tags in an XHTML document. | |
| 713 | + * | |
| 714 | + * Note: The tags will not be removed if they have any content. | |
| 715 | + * | |
| 716 | + * @param string $tagName the tag name, e.g., "p" | |
| 717 | + * | |
| 718 | + * @return void | |
| 719 | + */ | |
| 720 | + public function addUnprocessableHtmlTag($tagName) | |
| 721 | + { | |
| 722 | + $this->unprocessableHtmlTags[] = $tagName; | |
| 723 | + } | |
| 724 | + | |
| 725 | + /** | |
| 726 | + * Drops a tag from the removal list. | |
| 727 | + * | |
| 728 | + * @param string $tagName the tag name, e.g., "p" | |
| 729 | + * | |
| 730 | + * @return void | |
| 731 | + */ | |
| 732 | + public function removeUnprocessableHtmlTag($tagName) | |
| 733 | + { | |
| 734 | + $key = array_search($tagName, $this->unprocessableHtmlTags, true); | |
| 735 | + if ($key !== false) { | |
| 736 | + unset($this->unprocessableHtmlTags[$key]); | |
| 737 | + } | |
| 738 | + } | |
| 739 | + | |
| 740 | + /** | |
| 741 | + * Marks a media query type to keep. | |
| 742 | + * | |
| 743 | + * @param string $mediaName the media type name, e.g., "braille" | |
| 744 | + * | |
| 745 | + * @return void | |
| 746 | + */ | |
| 747 | + public function addAllowedMediaType($mediaName) | |
| 748 | + { | |
| 749 | + $this->allowedMediaTypes[$mediaName] = true; | |
| 750 | + } | |
| 751 | + | |
| 752 | + /** | |
| 753 | + * Drops a media query type from the allowed list. | |
| 754 | + * | |
| 755 | + * @param string $mediaName the tag name, e.g., "braille" | |
| 756 | + * | |
| 757 | + * @return void | |
| 758 | + */ | |
| 759 | + public function removeAllowedMediaType($mediaName) | |
| 760 | + { | |
| 761 | + if (isset($this->allowedMediaTypes[$mediaName])) { | |
| 762 | + unset($this->allowedMediaTypes[$mediaName]); | |
| 763 | + } | |
| 764 | + } | |
| 765 | + | |
| 766 | + /** | |
| 767 | + * Adds a selector to exclude nodes from emogrification. | |
| 768 | + * | |
| 769 | + * Any nodes that match the selector will not have their style altered. | |
| 770 | + * | |
| 771 | + * @param string $selector the selector to exclude, e.g., ".editor" | |
| 772 | + * | |
| 773 | + * @return void | |
| 774 | + */ | |
| 775 | + public function addExcludedSelector($selector) | |
| 776 | + { | |
| 777 | + $this->excludedSelectors[$selector] = true; | |
| 778 | + } | |
| 779 | + | |
| 780 | + /** | |
| 781 | + * No longer excludes the nodes matching this selector from emogrification. | |
| 782 | + * | |
| 783 | + * @param string $selector the selector to no longer exclude, e.g., ".editor" | |
| 784 | + * | |
| 785 | + * @return void | |
| 786 | + */ | |
| 787 | + public function removeExcludedSelector($selector) | |
| 788 | + { | |
| 789 | + if (isset($this->excludedSelectors[$selector])) { | |
| 790 | + unset($this->excludedSelectors[$selector]); | |
| 791 | + } | |
| 792 | + } | |
| 793 | + | |
| 794 | + /** | |
| 795 | + * This removes styles from your email that contain display:none. | |
| 796 | + * We need to look for display:none, but we need to do a case-insensitive search. Since DOMDocument only | |
| 797 | + * supports XPath 1.0, lower-case() isn't available to us. We've thus far only set attributes to lowercase, | |
| 798 | + * not attribute values. Consequently, we need to translate() the letters that would be in 'NONE' ("NOE") | |
| 799 | + * to lowercase. | |
| 800 | + * | |
| 801 | + * @param \DOMXPath $xPath | |
| 802 | + * | |
| 803 | + * @return void | |
| 804 | + */ | |
| 805 | + private function removeInvisibleNodes(\DOMXPath $xPath) | |
| 806 | + { | |
| 807 | + $nodesWithStyleDisplayNone = $xPath->query( | |
| 808 | + '//*[contains(translate(translate(@style," ",""),"NOE","noe"),"display:none")]' | |
| 809 | + ); | |
| 810 | + if ($nodesWithStyleDisplayNone->length === 0) { | |
| 811 | + return; | |
| 812 | + } | |
| 813 | + | |
| 814 | + // The checks on parentNode and is_callable below ensure that if we've deleted the parent node, | |
| 815 | + // we don't try to call removeChild on a nonexistent child node | |
| 816 | + /** @var \DOMNode $node */ | |
| 817 | + foreach ($nodesWithStyleDisplayNone as $node) { | |
| 818 | + if ($node->parentNode && is_callable([$node->parentNode, 'removeChild'])) { | |
| 819 | + $node->parentNode->removeChild($node); | |
| 820 | + } | |
| 821 | + } | |
| 822 | + } | |
| 823 | + | |
| 824 | + /** | |
| 825 | + * Normalizes the value of the "style" attribute and saves it. | |
| 826 | + * | |
| 827 | + * @param \DOMElement $node | |
| 828 | + * | |
| 829 | + * @return void | |
| 830 | + */ | |
| 831 | + private function normalizeStyleAttributes(\DOMElement $node) | |
| 832 | + { | |
| 833 | + $normalizedOriginalStyle = preg_replace_callback( | |
| 834 | + '/[A-z\\-]+(?=\\:)/S', | |
| 835 | + function (array $m) { | |
| 836 | + return strtolower($m[0]); | |
| 837 | + }, | |
| 838 | + $node->getAttribute('style') | |
| 839 | + ); | |
| 840 | + | |
| 841 | + // in order to not overwrite existing style attributes in the HTML, we | |
| 842 | + // have to save the original HTML styles | |
| 843 | + $nodePath = $node->getNodePath(); | |
| 844 | + if (!isset($this->styleAttributesForNodes[$nodePath])) { | |
| 845 | + $this->styleAttributesForNodes[$nodePath] = $this->parseCssDeclarationsBlock($normalizedOriginalStyle); | |
| 846 | + $this->visitedNodes[$nodePath] = $node; | |
| 847 | + } | |
| 848 | + | |
| 849 | + $node->setAttribute('style', $normalizedOriginalStyle); | |
| 850 | + } | |
| 851 | + | |
| 852 | + /** | |
| 853 | + * Merges styles from styles attributes and style nodes and applies them to the attribute nodes | |
| 854 | + * | |
| 855 | + * @return void | |
| 856 | + */ | |
| 857 | + private function fillStyleAttributesWithMergedStyles() | |
| 858 | + { | |
| 859 | + foreach ($this->styleAttributesForNodes as $nodePath => $styleAttributesForNode) { | |
| 860 | + $node = $this->visitedNodes[$nodePath]; | |
| 861 | + $currentStyleAttributes = $this->parseCssDeclarationsBlock($node->getAttribute('style')); | |
| 862 | + $node->setAttribute( | |
| 863 | + 'style', | |
| 864 | + $this->generateStyleStringFromDeclarationsArrays( | |
| 865 | + $currentStyleAttributes, | |
| 866 | + $styleAttributesForNode | |
| 867 | + ) | |
| 868 | + ); | |
| 869 | + } | |
| 870 | + } | |
| 871 | + | |
| 872 | + /** | |
| 873 | + * This method merges old or existing name/value array with new name/value array | |
| 874 | + * and then generates a string of the combined style suitable for placing inline. | |
| 875 | + * This becomes the single point for CSS string generation allowing for consistent | |
| 876 | + * CSS output no matter where the CSS originally came from. | |
| 877 | + * | |
| 878 | + * @param string[] $oldStyles | |
| 879 | + * @param string[] $newStyles | |
| 880 | + * | |
| 881 | + * @return string | |
| 882 | + */ | |
| 883 | + private function generateStyleStringFromDeclarationsArrays(array $oldStyles, array $newStyles) | |
| 884 | + { | |
| 885 | + $combinedStyles = array_merge($oldStyles, $newStyles); | |
| 886 | + $cacheKey = serialize($combinedStyles); | |
| 887 | + if (isset($this->caches[self::CACHE_KEY_COMBINED_STYLES][$cacheKey])) { | |
| 888 | + return $this->caches[self::CACHE_KEY_COMBINED_STYLES][$cacheKey]; | |
| 889 | + } | |
| 890 | + | |
| 891 | + foreach ($oldStyles as $attributeName => $attributeValue) { | |
| 892 | + if (!isset($newStyles[$attributeName])) { | |
| 893 | + continue; | |
| 894 | + } | |
| 895 | + | |
| 896 | + $newAttributeValue = $newStyles[$attributeName]; | |
| 897 | + if ($this->attributeValueIsImportant($attributeValue) | |
| 898 | + && !$this->attributeValueIsImportant($newAttributeValue) | |
| 899 | + ) { | |
| 900 | + $combinedStyles[$attributeName] = $attributeValue; | |
| 901 | + } | |
| 902 | + } | |
| 903 | + | |
| 904 | + $style = ''; | |
| 905 | + foreach ($combinedStyles as $attributeName => $attributeValue) { | |
| 906 | + $style .= strtolower(trim($attributeName)) . ': ' . trim($attributeValue) . '; '; | |
| 907 | + } | |
| 908 | + $trimmedStyle = rtrim($style); | |
| 909 | + | |
| 910 | + $this->caches[self::CACHE_KEY_COMBINED_STYLES][$cacheKey] = $trimmedStyle; | |
| 911 | + | |
| 912 | + return $trimmedStyle; | |
| 913 | + } | |
| 914 | + | |
| 915 | + /** | |
| 916 | + * Checks whether $attributeValue is marked as !important. | |
| 917 | + * | |
| 918 | + * @param string $attributeValue | |
| 919 | + * | |
| 920 | + * @return bool | |
| 921 | + */ | |
| 922 | + private function attributeValueIsImportant($attributeValue) | |
| 923 | + { | |
| 924 | + return strtolower(substr(trim($attributeValue), -10)) === '!important'; | |
| 925 | + } | |
| 926 | + | |
| 927 | + /** | |
| 928 | + * Applies $css to $xmlDocument, limited to the media queries that actually apply to the document. | |
| 929 | + * | |
| 930 | + * @param \DOMDocument $xmlDocument the document to match against | |
| 931 | + * @param \DOMXPath $xPath | |
| 932 | + * @param string $css a string of CSS | |
| 933 | + * | |
| 934 | + * @return void | |
| 935 | + */ | |
| 936 | + private function copyCssWithMediaToStyleNode(\DOMDocument $xmlDocument, \DOMXPath $xPath, $css) | |
| 937 | + { | |
| 938 | + if ($css === '') { | |
| 939 | + return; | |
| 940 | + } | |
| 941 | + | |
| 942 | + $mediaQueriesRelevantForDocument = []; | |
| 943 | + | |
| 944 | + foreach ($this->extractMediaQueriesFromCss($css) as $mediaQuery) { | |
| 945 | + foreach ($this->parseCssRules($mediaQuery['css']) as $selector) { | |
| 946 | + if ($this->existsMatchForCssSelector($xPath, $selector['selector'])) { | |
| 947 | + $mediaQueriesRelevantForDocument[] = $mediaQuery['query']; | |
| 948 | + break; | |
| 949 | + } | |
| 950 | + } | |
| 951 | + } | |
| 952 | + | |
| 953 | + $this->addStyleElementToDocument($xmlDocument, implode($mediaQueriesRelevantForDocument)); | |
| 954 | + } | |
| 955 | + | |
| 956 | + /** | |
| 957 | + * Extracts the media queries from $css while skipping empty media queries. | |
| 958 | + * | |
| 959 | + * @param string $css | |
| 960 | + * | |
| 961 | + * @return string[][] numeric array with string sub-arrays with the keys "css" and "query" | |
| 962 | + */ | |
| 963 | + private function extractMediaQueriesFromCss($css) | |
| 964 | + { | |
| 965 | + preg_match_all('/@media\\b[^{]*({((?:[^{}]+|(?1))*)})/', $css, $rawMediaQueries, PREG_SET_ORDER); | |
| 966 | + $parsedQueries = []; | |
| 967 | + | |
| 968 | + foreach ($rawMediaQueries as $mediaQuery) { | |
| 969 | + if ($mediaQuery[2] !== '') { | |
| 970 | + $parsedQueries[] = [ | |
| 971 | + 'css' => $mediaQuery[2], | |
| 972 | + 'query' => $mediaQuery[0], | |
| 973 | + ]; | |
| 974 | + } | |
| 975 | + } | |
| 976 | + | |
| 977 | + return $parsedQueries; | |
| 978 | + } | |
| 979 | + | |
| 980 | + /** | |
| 981 | + * Checks whether there is at least one matching element for $cssSelector. | |
| 982 | + * | |
| 983 | + * @param \DOMXPath $xPath | |
| 984 | + * @param string $cssSelector | |
| 985 | + * | |
| 986 | + * @return bool | |
| 987 | + */ | |
| 988 | + private function existsMatchForCssSelector(\DOMXPath $xPath, $cssSelector) | |
| 989 | + { | |
| 990 | + $nodesMatchingSelector = $xPath->query($this->translateCssToXpath($cssSelector)); | |
| 991 | + | |
| 992 | + return $nodesMatchingSelector !== false && $nodesMatchingSelector->length !== 0; | |
| 993 | + } | |
| 994 | + | |
| 995 | + /** | |
| 996 | + * Returns CSS content. | |
| 997 | + * | |
| 998 | + * @param \DOMXPath $xPath | |
| 999 | + * | |
| 1000 | + * @return string | |
| 1001 | + */ | |
| 1002 | + private function getCssFromAllStyleNodes(\DOMXPath $xPath) | |
| 1003 | + { | |
| 1004 | + $styleNodes = $xPath->query('//style'); | |
| 1005 | + | |
| 1006 | + if ($styleNodes === false) { | |
| 1007 | + return ''; | |
| 1008 | + } | |
| 1009 | + | |
| 1010 | + $css = ''; | |
| 1011 | + /** @var \DOMNode $styleNode */ | |
| 1012 | + foreach ($styleNodes as $styleNode) { | |
| 1013 | + $css .= "\n\n" . $styleNode->nodeValue; | |
| 1014 | + $styleNode->parentNode->removeChild($styleNode); | |
| 1015 | + } | |
| 1016 | + | |
| 1017 | + return $css; | |
| 1018 | + } | |
| 1019 | + | |
| 1020 | + /** | |
| 1021 | + * Adds a style element with $css to $document. | |
| 1022 | + * | |
| 1023 | + * This method is protected to allow overriding. | |
| 1024 | + * | |
| 1025 | + * @see https://github.com/jjriv/emogrifier/issues/103 | |
| 1026 | + * | |
| 1027 | + * @param \DOMDocument $document | |
| 1028 | + * @param string $css | |
| 1029 | + * | |
| 1030 | + * @return void | |
| 1031 | + */ | |
| 1032 | + protected function addStyleElementToDocument(\DOMDocument $document, $css) | |
| 1033 | + { | |
| 1034 | + $styleElement = $document->createElement('style', $css); | |
| 1035 | + $styleAttribute = $document->createAttribute('type'); | |
| 1036 | + $styleAttribute->value = 'text/css'; | |
| 1037 | + $styleElement->appendChild($styleAttribute); | |
| 1038 | + | |
| 1039 | + $head = $this->getOrCreateHeadElement($document); | |
| 1040 | + $head->appendChild($styleElement); | |
| 1041 | + } | |
| 1042 | + | |
| 1043 | + /** | |
| 1044 | + * Returns the existing or creates a new head element in $document. | |
| 1045 | + * | |
| 1046 | + * @param \DOMDocument $document | |
| 1047 | + * | |
| 1048 | + * @return \DOMNode the head element | |
| 1049 | + */ | |
| 1050 | + private function getOrCreateHeadElement(\DOMDocument $document) | |
| 1051 | + { | |
| 1052 | + $head = $document->getElementsByTagName('head')->item(0); | |
| 1053 | + | |
| 1054 | + if ($head === null) { | |
| 1055 | + $head = $document->createElement('head'); | |
| 1056 | + $html = $document->getElementsByTagName('html')->item(0); | |
| 1057 | + $html->insertBefore($head, $document->getElementsByTagName('body')->item(0)); | |
| 1058 | + } | |
| 1059 | + | |
| 1060 | + return $head; | |
| 1061 | + } | |
| 1062 | + | |
| 1063 | + /** | |
| 1064 | + * Splits input CSS code to an array where: | |
| 1065 | + * | |
| 1066 | + * - key "css" will be contains clean CSS code | |
| 1067 | + * - key "media" will be contains all valuable media queries | |
| 1068 | + * | |
| 1069 | + * Example: | |
| 1070 | + * | |
| 1071 | + * The CSS code | |
| 1072 | + * | |
| 1073 | + * "@import "file.css"; h1 { color:red; } @media { h1 {}} @media tv { h1 {}}" | |
| 1074 | + * | |
| 1075 | + * will be parsed into the following array: | |
| 1076 | + * | |
| 1077 | + * "css" => "h1 { color:red; }" | |
| 1078 | + * "media" => "@media { h1 {}}" | |
| 1079 | + * | |
| 1080 | + * @param string $css | |
| 1081 | + * | |
| 1082 | + * @return string[] | |
| 1083 | + */ | |
| 1084 | + private function splitCssAndMediaQuery($css) | |
| 1085 | + { | |
| 1086 | + $cssWithoutComments = preg_replace('/\\/\\*.*\\*\\//sU', '', $css); | |
| 1087 | + | |
| 1088 | + $mediaTypesExpression = ''; | |
| 1089 | + if (!empty($this->allowedMediaTypes)) { | |
| 1090 | + $mediaTypesExpression = '|' . implode('|', array_keys($this->allowedMediaTypes)); | |
| 1091 | + } | |
| 1092 | + | |
| 1093 | + $media = ''; | |
| 1094 | + $cssForAllowedMediaTypes = preg_replace_callback( | |
| 1095 | + '#@media\\s+(?:only\\s)?(?:[\\s{\\(]' . $mediaTypesExpression . ')\\s?[^{]+{.*}\\s*}\\s*#misU', | |
| 1096 | + function ($matches) use (&$media) { | |
| 1097 | + $media .= $matches[0]; | |
| 1098 | + }, | |
| 1099 | + $cssWithoutComments | |
| 1100 | + ); | |
| 1101 | + | |
| 1102 | + // filter the CSS | |
| 1103 | + $search = [ | |
| 1104 | + 'import directives' => '/^\\s*@import\\s[^;]+;/misU', | |
| 1105 | + 'remaining media enclosures' => '/^\\s*@media\\s[^{]+{(.*)}\\s*}\\s/misU', | |
| 1106 | + ]; | |
| 1107 | + | |
| 1108 | + $cleanedCss = preg_replace($search, '', $cssForAllowedMediaTypes); | |
| 1109 | + | |
| 1110 | + return ['css' => $cleanedCss, 'media' => $media]; | |
| 1111 | + } | |
| 1112 | + | |
| 1113 | + /** | |
| 1114 | + * Creates a DOMDocument instance with the current HTML. | |
| 1115 | + * | |
| 1116 | + * @return \DOMDocument | |
| 1117 | + */ | |
| 1118 | + private function createXmlDocument() | |
| 1119 | + { | |
| 1120 | + $xmlDocument = new \DOMDocument; | |
| 1121 | + $xmlDocument->encoding = 'UTF-8'; | |
| 1122 | + $xmlDocument->strictErrorChecking = false; | |
| 1123 | + $xmlDocument->formatOutput = true; | |
| 1124 | + $libXmlState = libxml_use_internal_errors(true); | |
| 1125 | + $xmlDocument->loadHTML($this->getUnifiedHtml()); | |
| 1126 | + libxml_clear_errors(); | |
| 1127 | + libxml_use_internal_errors($libXmlState); | |
| 1128 | + $xmlDocument->normalizeDocument(); | |
| 1129 | + | |
| 1130 | + return $xmlDocument; | |
| 1131 | + } | |
| 1132 | + | |
| 1133 | + /** | |
| 1134 | + * Returns the HTML with the unprocessable HTML tags removed and | |
| 1135 | + * with added document type and Content-Type meta tag if needed. | |
| 1136 | + * | |
| 1137 | + * @return string the unified HTML | |
| 1138 | + * | |
| 1139 | + * @throws \BadMethodCallException | |
| 1140 | + */ | |
| 1141 | + private function getUnifiedHtml() | |
| 1142 | + { | |
| 1143 | + $htmlWithoutUnprocessableTags = $this->removeUnprocessableTags($this->html); | |
| 1144 | + $htmlWithDocumentType = $this->ensureDocumentType($htmlWithoutUnprocessableTags); | |
| 1145 | + | |
| 1146 | + return $this->addContentTypeMetaTag($htmlWithDocumentType); | |
| 1147 | + } | |
| 1148 | + | |
| 1149 | + /** | |
| 1150 | + * Removes the unprocessable tags from $html (if this feature is enabled). | |
| 1151 | + * | |
| 1152 | + * @param string $html | |
| 1153 | + * | |
| 1154 | + * @return string the reworked HTML with the unprocessable tags removed | |
| 1155 | + */ | |
| 1156 | + private function removeUnprocessableTags($html) | |
| 1157 | + { | |
| 1158 | + if (empty($this->unprocessableHtmlTags)) { | |
| 1159 | + return $html; | |
| 1160 | + } | |
| 1161 | + | |
| 1162 | + $unprocessableHtmlTags = implode('|', $this->unprocessableHtmlTags); | |
| 1163 | + | |
| 1164 | + return preg_replace( | |
| 1165 | + '/<\\/?(' . $unprocessableHtmlTags . ')[^>]*>/i', | |
| 1166 | + '', | |
| 1167 | + $html | |
| 1168 | + ); | |
| 1169 | + } | |
| 1170 | + | |
| 1171 | + /** | |
| 1172 | + * Makes sure that the passed HTML has a document type. | |
| 1173 | + * | |
| 1174 | + * @param string $html | |
| 1175 | + * | |
| 1176 | + * @return string HTML with document type | |
| 1177 | + */ | |
| 1178 | + private function ensureDocumentType($html) | |
| 1179 | + { | |
| 1180 | + $hasDocumentType = stripos($html, '<!DOCTYPE') !== false; | |
| 1181 | + if ($hasDocumentType) { | |
| 1182 | + return $html; | |
| 1183 | + } | |
| 1184 | + | |
| 1185 | + return self::DEFAULT_DOCUMENT_TYPE . $html; | |
| 1186 | + } | |
| 1187 | + | |
| 1188 | + /** | |
| 1189 | + * Adds a Content-Type meta tag for the charset. | |
| 1190 | + * | |
| 1191 | + * @param string $html | |
| 1192 | + * | |
| 1193 | + * @return string the HTML with the meta tag added | |
| 1194 | + */ | |
| 1195 | + private function addContentTypeMetaTag($html) | |
| 1196 | + { | |
| 1197 | + $hasContentTypeMetaTag = stristr($html, 'Content-Type') !== false; | |
| 1198 | + if ($hasContentTypeMetaTag) { | |
| 1199 | + return $html; | |
| 1200 | + } | |
| 1201 | + | |
| 1202 | + // We are trying to insert the meta tag to the right spot in the DOM. | |
| 1203 | + // If we just prepended it to the HTML, we would lose attributes set to the HTML tag. | |
| 1204 | + $hasHeadTag = stripos($html, '<head') !== false; | |
| 1205 | + $hasHtmlTag = stripos($html, '<html') !== false; | |
| 1206 | + | |
| 1207 | + if ($hasHeadTag) { | |
| 1208 | + $reworkedHtml = preg_replace('/<head(.*?)>/i', '<head$1>' . self::CONTENT_TYPE_META_TAG, $html); | |
| 1209 | + } elseif ($hasHtmlTag) { | |
| 1210 | + $reworkedHtml = preg_replace( | |
| 1211 | + '/<html(.*?)>/i', | |
| 1212 | + '<html$1><head>' . self::CONTENT_TYPE_META_TAG . '</head>', | |
| 1213 | + $html | |
| 1214 | + ); | |
| 1215 | + } else { | |
| 1216 | + $reworkedHtml = self::CONTENT_TYPE_META_TAG . $html; | |
| 1217 | + } | |
| 1218 | + | |
| 1219 | + return $reworkedHtml; | |
| 1220 | + } | |
| 1221 | + | |
| 1222 | + /** | |
| 1223 | + * @param string[] $a | |
| 1224 | + * @param string[] $b | |
| 1225 | + * | |
| 1226 | + * @return int | |
| 1227 | + */ | |
| 1228 | + private function sortBySelectorPrecedence(array $a, array $b) | |
| 1229 | + { | |
| 1230 | + $precedenceA = $this->getCssSelectorPrecedence($a['selector']); | |
| 1231 | + $precedenceB = $this->getCssSelectorPrecedence($b['selector']); | |
| 1232 | + | |
| 1233 | + // We want these sorted in ascending order so selectors with lesser precedence get processed first and | |
| 1234 | + // selectors with greater precedence get sorted last. | |
| 1235 | + $precedenceForEquals = ($a['line'] < $b['line'] ? -1 : 1); | |
| 1236 | + $precedenceForNotEquals = ($precedenceA < $precedenceB ? -1 : 1); | |
| 1237 | + return ($precedenceA === $precedenceB) ? $precedenceForEquals : $precedenceForNotEquals; | |
| 1238 | + } | |
| 1239 | + | |
| 1240 | + /** | |
| 1241 | + * @param string $selector | |
| 1242 | + * | |
| 1243 | + * @return int | |
| 1244 | + */ | |
| 1245 | + private function getCssSelectorPrecedence($selector) | |
| 1246 | + { | |
| 1247 | + $selectorKey = md5($selector); | |
| 1248 | + if (!isset($this->caches[self::CACHE_KEY_SELECTOR][$selectorKey])) { | |
| 1249 | + $precedence = 0; | |
| 1250 | + $value = 100; | |
| 1251 | + // ids: worth 100, classes: worth 10, elements: worth 1 | |
| 1252 | + $search = ['\\#','\\.','']; | |
| 1253 | + | |
| 1254 | + foreach ($search as $s) { | |
| 1255 | + if (trim($selector) === '') { | |
| 1256 | + break; | |
| 1257 | + } | |
| 1258 | + $number = 0; | |
| 1259 | + $selector = preg_replace('/' . $s . '\\w+/', '', $selector, -1, $number); | |
| 1260 | + $precedence += ($value * $number); | |
| 1261 | + $value /= 10; | |
| 1262 | + } | |
| 1263 | + $this->caches[self::CACHE_KEY_SELECTOR][$selectorKey] = $precedence; | |
| 1264 | + } | |
| 1265 | + | |
| 1266 | + return $this->caches[self::CACHE_KEY_SELECTOR][$selectorKey]; | |
| 1267 | + } | |
| 1268 | + | |
| 1269 | + /** | |
| 1270 | + * Maps a CSS selector to an XPath query string. | |
| 1271 | + * | |
| 1272 | + * @see http://plasmasturm.org/log/444/ | |
| 1273 | + * | |
| 1274 | + * @param string $cssSelector a CSS selector | |
| 1275 | + * | |
| 1276 | + * @return string the corresponding XPath selector | |
| 1277 | + */ | |
| 1278 | + private function translateCssToXpath($cssSelector) | |
| 1279 | + { | |
| 1280 | + $paddedSelector = ' ' . $cssSelector . ' '; | |
| 1281 | + $lowercasePaddedSelector = preg_replace_callback( | |
| 1282 | + '/\\s+\\w+\\s+/', | |
| 1283 | + function (array $matches) { | |
| 1284 | + return strtolower($matches[0]); | |
| 1285 | + }, | |
| 1286 | + $paddedSelector | |
| 1287 | + ); | |
| 1288 | + $trimmedLowercaseSelector = trim($lowercasePaddedSelector); | |
| 1289 | + $xPathKey = md5($trimmedLowercaseSelector); | |
| 1290 | + if (!isset($this->caches[self::CACHE_KEY_XPATH][$xPathKey])) { | |
| 1291 | + $roughXpath = '//' . preg_replace( | |
| 1292 | + array_keys($this->xPathRules), | |
| 1293 | + $this->xPathRules, | |
| 1294 | + $trimmedLowercaseSelector | |
| 1295 | + ); | |
| 1296 | + $xPathWithIdAttributeMatchers = preg_replace_callback( | |
| 1297 | + self::ID_ATTRIBUTE_MATCHER, | |
| 1298 | + [$this, 'matchIdAttributes'], | |
| 1299 | + $roughXpath | |
| 1300 | + ); | |
| 1301 | + $xPathWithIdAttributeAndClassMatchers = preg_replace_callback( | |
| 1302 | + self::CLASS_ATTRIBUTE_MATCHER, | |
| 1303 | + [$this, 'matchClassAttributes'], | |
| 1304 | + $xPathWithIdAttributeMatchers | |
| 1305 | + ); | |
| 1306 | + | |
| 1307 | + // Advanced selectors are going to require a bit more advanced emogrification. | |
| 1308 | + // When we required PHP 5.3, we could do this with closures. | |
| 1309 | + $xPathWithIdAttributeAndClassMatchers = preg_replace_callback( | |
| 1310 | + '/([^\\/]+):nth-child\\(\\s*(odd|even|[+\\-]?\\d|[+\\-]?\\d?n(\\s*[+\\-]\\s*\\d)?)\\s*\\)/i', | |
| 1311 | + [$this, 'translateNthChild'], | |
| 1312 | + $xPathWithIdAttributeAndClassMatchers | |
| 1313 | + ); | |
| 1314 | + $finalXpath = preg_replace_callback( | |
| 1315 | + '/([^\\/]+):nth-of-type\\(\s*(odd|even|[+\\-]?\\d|[+\\-]?\\d?n(\\s*[+\\-]\\s*\\d)?)\\s*\\)/i', | |
| 1316 | + [$this, 'translateNthOfType'], | |
| 1317 | + $xPathWithIdAttributeAndClassMatchers | |
| 1318 | + ); | |
| 1319 | + | |
| 1320 | + $this->caches[self::CACHE_KEY_SELECTOR][$xPathKey] = $finalXpath; | |
| 1321 | + } | |
| 1322 | + return $this->caches[self::CACHE_KEY_SELECTOR][$xPathKey]; | |
| 1323 | + } | |
| 1324 | + | |
| 1325 | + /** | |
| 1326 | + * @param string[] $match | |
| 1327 | + * | |
| 1328 | + * @return string | |
| 1329 | + */ | |
| 1330 | + private function matchIdAttributes(array $match) | |
| 1331 | + { | |
| 1332 | + return ($match[1] !== '' ? $match[1] : '*') . '[@id="' . $match[2] . '"]'; | |
| 1333 | + } | |
| 1334 | + | |
| 1335 | + /** | |
| 1336 | + * @param string[] $match | |
| 1337 | + * | |
| 1338 | + * @return string | |
| 1339 | + */ | |
| 1340 | + private function matchClassAttributes(array $match) | |
| 1341 | + { | |
| 1342 | + return ($match[1] !== '' ? $match[1] : '*') . '[contains(concat(" ",@class," "),concat(" ","' . | |
| 1343 | + implode( | |
| 1344 | + '"," "))][contains(concat(" ",@class," "),concat(" ","', | |
| 1345 | + explode('.', substr($match[2], 1)) | |
| 1346 | + ) . '"," "))]'; | |
| 1347 | + } | |
| 1348 | + | |
| 1349 | + /** | |
| 1350 | + * @param string[] $match | |
| 1351 | + * | |
| 1352 | + * @return string | |
| 1353 | + */ | |
| 1354 | + private function translateNthChild(array $match) | |
| 1355 | + { | |
| 1356 | + $parseResult = $this->parseNth($match); | |
| 1357 | + | |
| 1358 | + if (isset($parseResult[self::MULTIPLIER])) { | |
| 1359 | + if ($parseResult[self::MULTIPLIER] < 0) { | |
| 1360 | + $parseResult[self::MULTIPLIER] = abs($parseResult[self::MULTIPLIER]); | |
| 1361 | + $xPathExpression = sprintf( | |
| 1362 | + '*[(last() - position()) mod %u = %u]/self::%s', | |
| 1363 | + $parseResult[self::MULTIPLIER], | |
| 1364 | + $parseResult[self::INDEX], | |
| 1365 | + $match[1] | |
| 1366 | + ); | |
| 1367 | + } else { | |
| 1368 | + $xPathExpression = sprintf( | |
| 1369 | + '*[position() mod %u = %u]/self::%s', | |
| 1370 | + $parseResult[self::MULTIPLIER], | |
| 1371 | + $parseResult[self::INDEX], | |
| 1372 | + $match[1] | |
| 1373 | + ); | |
| 1374 | + } | |
| 1375 | + } else { | |
| 1376 | + $xPathExpression = sprintf('*[%u]/self::%s', $parseResult[self::INDEX], $match[1]); | |
| 1377 | + } | |
| 1378 | + | |
| 1379 | + return $xPathExpression; | |
| 1380 | + } | |
| 1381 | + | |
| 1382 | + /** | |
| 1383 | + * @param string[] $match | |
| 1384 | + * | |
| 1385 | + * @return string | |
| 1386 | + */ | |
| 1387 | + private function translateNthOfType(array $match) | |
| 1388 | + { | |
| 1389 | + $parseResult = $this->parseNth($match); | |
| 1390 | + | |
| 1391 | + if (isset($parseResult[self::MULTIPLIER])) { | |
| 1392 | + if ($parseResult[self::MULTIPLIER] < 0) { | |
| 1393 | + $parseResult[self::MULTIPLIER] = abs($parseResult[self::MULTIPLIER]); | |
| 1394 | + $xPathExpression = sprintf( | |
| 1395 | + '%s[(last() - position()) mod %u = %u]', | |
| 1396 | + $match[1], | |
| 1397 | + $parseResult[self::MULTIPLIER], | |
| 1398 | + $parseResult[self::INDEX] | |
| 1399 | + ); | |
| 1400 | + } else { | |
| 1401 | + $xPathExpression = sprintf( | |
| 1402 | + '%s[position() mod %u = %u]', | |
| 1403 | + $match[1], | |
| 1404 | + $parseResult[self::MULTIPLIER], | |
| 1405 | + $parseResult[self::INDEX] | |
| 1406 | + ); | |
| 1407 | + } | |
| 1408 | + } else { | |
| 1409 | + $xPathExpression = sprintf('%s[%u]', $match[1], $parseResult[self::INDEX]); | |
| 1410 | + } | |
| 1411 | + | |
| 1412 | + return $xPathExpression; | |
| 1413 | + } | |
| 1414 | + | |
| 1415 | + /** | |
| 1416 | + * @param string[] $match | |
| 1417 | + * | |
| 1418 | + * @return int[] | |
| 1419 | + */ | |
| 1420 | + private function parseNth(array $match) | |
| 1421 | + { | |
| 1422 | + if (in_array(strtolower($match[2]), ['even', 'odd'], true)) { | |
| 1423 | + // we have "even" or "odd" | |
| 1424 | + $index = strtolower($match[2]) === 'even' ? 0 : 1; | |
| 1425 | + return [self::MULTIPLIER => 2, self::INDEX => $index]; | |
| 1426 | + } | |
| 1427 | + if (stripos($match[2], 'n') === false) { | |
| 1428 | + // if there is a multiplier | |
| 1429 | + $index = (int) str_replace(' ', '', $match[2]); | |
| 1430 | + return [self::INDEX => $index]; | |
| 1431 | + } | |
| 1432 | + | |
| 1433 | + if (isset($match[3])) { | |
| 1434 | + $multipleTerm = str_replace($match[3], '', $match[2]); | |
| 1435 | + $index = (int) str_replace(' ', '', $match[3]); | |
| 1436 | + } else { | |
| 1437 | + $multipleTerm = $match[2]; | |
| 1438 | + $index = 0; | |
| 1439 | + } | |
| 1440 | + | |
| 1441 | + $multiplier = str_ireplace('n', '', $multipleTerm); | |
| 1442 | + | |
| 1443 | + if ($multiplier === '') { | |
| 1444 | + $multiplier = 1; | |
| 1445 | + } elseif ($multiplier === '0') { | |
| 1446 | + return [self::INDEX => $index]; | |
| 1447 | + } else { | |
| 1448 | + $multiplier = (int) $multiplier; | |
| 1449 | + } | |
| 1450 | + | |
| 1451 | + while ($index < 0) { | |
| 1452 | + $index += abs($multiplier); | |
| 1453 | + } | |
| 1454 | + | |
| 1455 | + return [self::MULTIPLIER => $multiplier, self::INDEX => $index]; | |
| 1456 | + } | |
| 1457 | + | |
| 1458 | + /** | |
| 1459 | + * Parses a CSS declaration block into property name/value pairs. | |
| 1460 | + * | |
| 1461 | + * Example: | |
| 1462 | + * | |
| 1463 | + * The declaration block | |
| 1464 | + * | |
| 1465 | + * "color: #000; font-weight: bold;" | |
| 1466 | + * | |
| 1467 | + * will be parsed into the following array: | |
| 1468 | + * | |
| 1469 | + * "color" => "#000" | |
| 1470 | + * "font-weight" => "bold" | |
| 1471 | + * | |
| 1472 | + * @param string $cssDeclarationsBlock the CSS declarations block without the curly braces, may be empty | |
| 1473 | + * | |
| 1474 | + * @return string[] | |
| 1475 | + * the CSS declarations with the property names as array keys and the property values as array values | |
| 1476 | + */ | |
| 1477 | + private function parseCssDeclarationsBlock($cssDeclarationsBlock) | |
| 1478 | + { | |
| 1479 | + if (isset($this->caches[self::CACHE_KEY_CSS_DECLARATIONS_BLOCK][$cssDeclarationsBlock])) { | |
| 1480 | + return $this->caches[self::CACHE_KEY_CSS_DECLARATIONS_BLOCK][$cssDeclarationsBlock]; | |
| 1481 | + } | |
| 1482 | + | |
| 1483 | + $properties = []; | |
| 1484 | + $declarations = preg_split('/;(?!base64|charset)/', $cssDeclarationsBlock); | |
| 1485 | + | |
| 1486 | + foreach ($declarations as $declaration) { | |
| 1487 | + $matches = []; | |
| 1488 | + if (!preg_match('/^([A-Za-z\\-]+)\\s*:\\s*(.+)$/', trim($declaration), $matches)) { | |
| 1489 | + continue; | |
| 1490 | + } | |
| 1491 | + | |
| 1492 | + $propertyName = strtolower($matches[1]); | |
| 1493 | + $propertyValue = $matches[2]; | |
| 1494 | + $properties[$propertyName] = $propertyValue; | |
| 1495 | + } | |
| 1496 | + $this->caches[self::CACHE_KEY_CSS_DECLARATIONS_BLOCK][$cssDeclarationsBlock] = $properties; | |
| 1497 | + | |
| 1498 | + return $properties; | |
| 1499 | + } | |
| 1500 | + | |
| 1501 | + /** | |
| 1502 | + * Find the nodes that are not to be emogrified. | |
| 1503 | + * | |
| 1504 | + * @param \DOMXPath $xPath | |
| 1505 | + * | |
| 1506 | + * @return \DOMElement[] | |
| 1507 | + */ | |
| 1508 | + private function getNodesToExclude(\DOMXPath $xPath) | |
| 1509 | + { | |
| 1510 | + $excludedNodes = []; | |
| 1511 | + foreach (array_keys($this->excludedSelectors) as $selectorToExclude) { | |
| 1512 | + foreach ($xPath->query($this->translateCssToXpath($selectorToExclude)) as $node) { | |
| 1513 | + $excludedNodes[] = $node; | |
| 1514 | + } | |
| 1515 | + } | |
| 1516 | + | |
| 1517 | + return $excludedNodes; | |
| 1518 | + } | |
| 1519 | + | |
| 1520 | + /** | |
| 1521 | + * Handles invalid xPath expression warnings, generated by process() method, | |
| 1522 | + * during querying \DOMDocument and trigger \InvalidArgumentException | |
| 1523 | + * with invalid selector. | |
| 1524 | + * | |
| 1525 | + * @param int $type | |
| 1526 | + * @param string $message | |
| 1527 | + * @param string $file | |
| 1528 | + * @param int $line | |
| 1529 | + * @param array $context | |
| 1530 | + * | |
| 1531 | + * @return bool always false | |
| 1532 | + * | |
| 1533 | + * @throws \InvalidArgumentException | |
| 1534 | + */ | |
| 1535 | + public function handleXpathError($type, $message, $file, $line, array $context) | |
| 1536 | + { | |
| 1537 | + if ($this->debug && $type === E_WARNING && isset($context['cssRule']['selector'])) { | |
| 1538 | + throw new \InvalidArgumentException( | |
| 1539 | + sprintf( | |
| 1540 | + '%s in selector >> %s << in %s on line %s', | |
| 1541 | + $message, | |
| 1542 | + $context['cssRule']['selector'], | |
| 1543 | + $file, | |
| 1544 | + $line | |
| 1545 | + ) | |
| 1546 | + ); | |
| 1547 | + } | |
| 1548 | + | |
| 1549 | + // the normal error handling continues when handler return false | |
| 1550 | + return false; | |
| 1551 | + } | |
| 1552 | + | |
| 1553 | + /** | |
| 1554 | + * Sets the debug mode. | |
| 1555 | + * | |
| 1556 | + * @param bool $debug set to true to enable debug mode | |
| 1557 | + * | |
| 1558 | + * @return void | |
| 1559 | + */ | |
| 1560 | + public function setDebug($debug) | |
| 1561 | + { | |
| 1562 | + $this->debug = $debug; | |
| 1563 | + } | |
| 1564 | +} | |
