File: /var/www/vhosts/uyarreklam.com.tr/httpdocs/HtmlProcessor.tar
AbstractHtmlProcessor.php 0000644 00000035420 15155373351 0011561 0 ustar 00 <?php
declare(strict_types=1);
namespace Pelago\Emogrifier\HtmlProcessor;
/**
* Base class for HTML processor that e.g., can remove, add or modify nodes or attributes.
*
* The "vanilla" subclass is the HtmlNormalizer.
*
* @psalm-consistent-constructor
*/
abstract class AbstractHtmlProcessor
{
/**
* @var string
*/
protected const DEFAULT_DOCUMENT_TYPE = '<!DOCTYPE html>';
/**
* @var string
*/
protected const CONTENT_TYPE_META_TAG = '<meta http-equiv="Content-Type" content="text/html; charset=utf-8">';
/**
* @var string Regular expression part to match tag names that PHP's DOMDocument implementation is not aware are
* self-closing. These are mostly HTML5 elements, but for completeness <command> (obsolete) and <keygen>
* (deprecated) are also included.
*
* @see https://bugs.php.net/bug.php?id=73175
*/
protected const PHP_UNRECOGNIZED_VOID_TAGNAME_MATCHER = '(?:command|embed|keygen|source|track|wbr)';
/**
* Regular expression part to match tag names that may appear before the start of the `<body>` element. A start tag
* for any other element would implicitly start the `<body>` element due to tag omission rules.
*
* @var string
*/
protected const TAGNAME_ALLOWED_BEFORE_BODY_MATCHER
= '(?:html|head|base|command|link|meta|noscript|script|style|template|title)';
/**
* regular expression pattern to match an HTML comment, including delimiters and modifiers
*
* @var string
*/
protected const HTML_COMMENT_PATTERN = '/<!--[^-]*+(?:-(?!->)[^-]*+)*+(?:-->|$)/';
/**
* regular expression pattern to match an HTML `<template>` element, including delimiters and modifiers
*
* @var string
*/
protected const HTML_TEMPLATE_ELEMENT_PATTERN
= '%<template[\\s>][^<]*+(?:<(?!/template>)[^<]*+)*+(?:</template>|$)%i';
/**
* @var ?\DOMDocument
*/
protected $domDocument = null;
/**
* @var ?\DOMXPath
*/
private $xPath = null;
/**
* The constructor.
*
* Please use `::fromHtml` or `::fromDomDocument` instead.
*/
private function __construct()
{
}
/**
* Builds a new instance from the given HTML.
*
* @param string $unprocessedHtml raw HTML, must be UTF-encoded, must not be empty
*
* @return static
*
* @throws \InvalidArgumentException if $unprocessedHtml is anything other than a non-empty string
*/
public static function fromHtml(string $unprocessedHtml): self
{
if ($unprocessedHtml === '') {
throw new \InvalidArgumentException('The provided HTML must not be empty.', 1515763647);
}
$instance = new static();
$instance->setHtml($unprocessedHtml);
return $instance;
}
/**
* Builds a new instance from the given DOM document.
*
* @param \DOMDocument $document a DOM document returned by getDomDocument() of another instance
*
* @return static
*/
public static function fromDomDocument(\DOMDocument $document): self
{
$instance = new static();
$instance->setDomDocument($document);
return $instance;
}
/**
* Sets the HTML to process.
*
* @param string $html the HTML to process, must be UTF-8-encoded
*/
private function setHtml(string $html): void
{
$this->createUnifiedDomDocument($html);
}
/**
* Provides access to the internal DOMDocument representation of the HTML in its current state.
*
* @return \DOMDocument
*
* @throws \UnexpectedValueException
*/
public function getDomDocument(): \DOMDocument
{
if (!$this->domDocument instanceof \DOMDocument) {
$message = self::class . '::setDomDocument() has not yet been called on ' . static::class;
throw new \UnexpectedValueException($message, 1570472239);
}
return $this->domDocument;
}
/**
* @param \DOMDocument $domDocument
*/
private function setDomDocument(\DOMDocument $domDocument): void
{
$this->domDocument = $domDocument;
$this->xPath = new \DOMXPath($this->domDocument);
}
/**
* @return \DOMXPath
*
* @throws \UnexpectedValueException
*/
protected function getXPath(): \DOMXPath
{
if (!$this->xPath instanceof \DOMXPath) {
$message = self::class . '::setDomDocument() has not yet been called on ' . static::class;
throw new \UnexpectedValueException($message, 1617819086);
}
return $this->xPath;
}
/**
* Renders the normalized and processed HTML.
*
* @return string
*/
public function render(): string
{
$htmlWithPossibleErroneousClosingTags = $this->getDomDocument()->saveHTML();
return $this->removeSelfClosingTagsClosingTags($htmlWithPossibleErroneousClosingTags);
}
/**
* Renders the content of the BODY element of the normalized and processed HTML.
*
* @return string
*/
public function renderBodyContent(): string
{
$htmlWithPossibleErroneousClosingTags = $this->getDomDocument()->saveHTML($this->getBodyElement());
$bodyNodeHtml = $this->removeSelfClosingTagsClosingTags($htmlWithPossibleErroneousClosingTags);
return \preg_replace('%</?+body(?:\\s[^>]*+)?+>%', '', $bodyNodeHtml);
}
/**
* Eliminates any invalid closing tags for void elements from the given HTML.
*
* @param string $html
*
* @return string
*/
private function removeSelfClosingTagsClosingTags(string $html): string
{
return \preg_replace('%</' . self::PHP_UNRECOGNIZED_VOID_TAGNAME_MATCHER . '>%', '', $html);
}
/**
* Returns the BODY element.
*
* This method assumes that there always is a BODY element.
*
* @return \DOMElement
*
* @throws \RuntimeException
*/
private function getBodyElement(): \DOMElement
{
$node = $this->getDomDocument()->getElementsByTagName('body')->item(0);
if (!$node instanceof \DOMElement) {
throw new \RuntimeException('There is no body element.', 1617922607);
}
return $node;
}
/**
* Creates a DOM document from the given HTML and stores it in $this->domDocument.
*
* The DOM document will always have a BODY element and a document type.
*
* @param string $html
*/
private function createUnifiedDomDocument(string $html): void
{
$this->createRawDomDocument($html);
$this->ensureExistenceOfBodyElement();
}
/**
* Creates a DOMDocument instance from the given HTML and stores it in $this->domDocument.
*
* @param string $html
*/
private function createRawDomDocument(string $html): void
{
$domDocument = new \DOMDocument();
$domDocument->strictErrorChecking = false;
$domDocument->formatOutput = true;
$libXmlState = \libxml_use_internal_errors(true);
$domDocument->loadHTML($this->prepareHtmlForDomConversion($html));
\libxml_clear_errors();
\libxml_use_internal_errors($libXmlState);
$this->setDomDocument($domDocument);
}
/**
* Returns the HTML with added document type, Content-Type meta tag, and self-closing slashes, if needed,
* ensuring that the HTML will be good for creating a DOM document from it.
*
* @param string $html
*
* @return string the unified HTML
*/
private function prepareHtmlForDomConversion(string $html): string
{
$htmlWithSelfClosingSlashes = $this->ensurePhpUnrecognizedSelfClosingTagsAreXml($html);
$htmlWithDocumentType = $this->ensureDocumentType($htmlWithSelfClosingSlashes);
return $this->addContentTypeMetaTag($htmlWithDocumentType);
}
/**
* Makes sure that the passed HTML has a document type, with lowercase "html".
*
* @param string $html
*
* @return string HTML with document type
*/
private function ensureDocumentType(string $html): string
{
$hasDocumentType = \stripos($html, '<!DOCTYPE') !== false;
if ($hasDocumentType) {
return $this->normalizeDocumentType($html);
}
return self::DEFAULT_DOCUMENT_TYPE . $html;
}
/**
* Makes sure the document type in the passed HTML has lowercase "html".
*
* @param string $html
*
* @return string HTML with normalized document type
*/
private function normalizeDocumentType(string $html): string
{
// Limit to replacing the first occurrence: as an optimization; and in case an example exists as unescaped text.
return \preg_replace(
'/<!DOCTYPE\\s++html(?=[\\s>])/i',
'<!DOCTYPE html',
$html,
1
);
}
/**
* Adds a Content-Type meta tag for the charset.
*
* This method also ensures that there is a HEAD element.
*
* @param string $html
*
* @return string the HTML with the meta tag added
*/
private function addContentTypeMetaTag(string $html): string
{
if ($this->hasContentTypeMetaTagInHead($html)) {
return $html;
}
// We are trying to insert the meta tag to the right spot in the DOM.
// If we just prepended it to the HTML, we would lose attributes set to the HTML tag.
$hasHeadTag = \preg_match('/<head[\\s>]/i', $html);
$hasHtmlTag = \stripos($html, '<html') !== false;
if ($hasHeadTag) {
$reworkedHtml = \preg_replace(
'/<head(?=[\\s>])([^>]*+)>/i',
'<head$1>' . self::CONTENT_TYPE_META_TAG,
$html
);
} elseif ($hasHtmlTag) {
$reworkedHtml = \preg_replace(
'/<html(.*?)>/is',
'<html$1><head>' . self::CONTENT_TYPE_META_TAG . '</head>',
$html
);
} else {
$reworkedHtml = self::CONTENT_TYPE_META_TAG . $html;
}
return $reworkedHtml;
}
/**
* Tests whether the given HTML has a valid `Content-Type` metadata element within the `<head>` element. Due to tag
* omission rules, HTML parsers are expected to end the `<head>` element and start the `<body>` element upon
* encountering a start tag for any element which is permitted only within the `<body>`.
*
* @param string $html
*
* @return bool
*/
private function hasContentTypeMetaTagInHead(string $html): bool
{
\preg_match('%^.*?(?=<meta(?=\\s)[^>]*\\shttp-equiv=(["\']?+)Content-Type\\g{-1}[\\s/>])%is', $html, $matches);
if (isset($matches[0])) {
$htmlBefore = $matches[0];
try {
$hasContentTypeMetaTagInHead = !$this->hasEndOfHeadElement($htmlBefore);
} catch (\RuntimeException $exception) {
// If something unexpected occurs, assume the `Content-Type` that was found is valid.
\trigger_error($exception->getMessage());
$hasContentTypeMetaTagInHead = true;
}
} else {
$hasContentTypeMetaTagInHead = false;
}
return $hasContentTypeMetaTagInHead;
}
/**
* Tests whether the `<head>` element ends within the given HTML. Due to tag omission rules, HTML parsers are
* expected to end the `<head>` element and start the `<body>` element upon encountering a start tag for any element
* which is permitted only within the `<body>`.
*
* @param string $html
*
* @return bool
*
* @throws \RuntimeException
*/
private function hasEndOfHeadElement(string $html): bool
{
$headEndTagMatchCount
= \preg_match('%<(?!' . self::TAGNAME_ALLOWED_BEFORE_BODY_MATCHER . '[\\s/>])\\w|</head>%i', $html);
if (\is_int($headEndTagMatchCount) && $headEndTagMatchCount > 0) {
// An exception to the implicit end of the `<head>` is any content within a `<template>` element, as well in
// comments. As an optimization, this is only checked for if a potential `<head>` end tag is found.
$htmlWithoutCommentsOrTemplates = $this->removeHtmlTemplateElements($this->removeHtmlComments($html));
$hasEndOfHeadElement = $htmlWithoutCommentsOrTemplates === $html
|| $this->hasEndOfHeadElement($htmlWithoutCommentsOrTemplates);
} else {
$hasEndOfHeadElement = false;
}
return $hasEndOfHeadElement;
}
/**
* Removes comments from the given HTML, including any which are unterminated, for which the remainder of the string
* is removed.
*
* @param string $html
*
* @return string
*
* @throws \RuntimeException
*/
private function removeHtmlComments(string $html): string
{
$result = \preg_replace(self::HTML_COMMENT_PATTERN, '', $html);
if (!\is_string($result)) {
throw new \RuntimeException('Internal PCRE error', 1616521475);
}
return $result;
}
/**
* Removes `<template>` elements from the given HTML, including any without an end tag, for which the remainder of
* the string is removed.
*
* @param string $html
*
* @return string
*
* @throws \RuntimeException
*/
private function removeHtmlTemplateElements(string $html): string
{
$result = \preg_replace(self::HTML_TEMPLATE_ELEMENT_PATTERN, '', $html);
if (!\is_string($result)) {
throw new \RuntimeException('Internal PCRE error', 1616519652);
}
return $result;
}
/**
* Makes sure that any self-closing tags not recognized as such by PHP's DOMDocument implementation have a
* self-closing slash.
*
* @param string $html
*
* @return string HTML with problematic tags converted.
*/
private function ensurePhpUnrecognizedSelfClosingTagsAreXml(string $html): string
{
return \preg_replace(
'%<' . self::PHP_UNRECOGNIZED_VOID_TAGNAME_MATCHER . '\\b[^>]*+(?<!/)(?=>)%',
'$0/',
$html
);
}
/**
* Checks that $this->domDocument has a BODY element and adds it if it is missing.
*
* @throws \UnexpectedValueException
*/
private function ensureExistenceOfBodyElement(): void
{
if ($this->getDomDocument()->getElementsByTagName('body')->item(0) instanceof \DOMElement) {
return;
}
$htmlElement = $this->getDomDocument()->getElementsByTagName('html')->item(0);
if (!$htmlElement instanceof \DOMElement) {
throw new \UnexpectedValueException('There is no HTML element although there should be one.', 1569930853);
}
$htmlElement->appendChild($this->getDomDocument()->createElement('body'));
}
}
CssToAttributeConverter.php 0000644 00000024101 15155373351 0012072 0 ustar 00 <?php
declare(strict_types=1);
namespace Pelago\Emogrifier\HtmlProcessor;
/**
* This HtmlProcessor can convert style HTML attributes to the corresponding other visual HTML attributes,
* e.g. it converts style="width: 100px" to width="100".
*
* It will only add attributes, but leaves the style attribute untouched.
*
* To trigger the conversion, call the convertCssToVisualAttributes method.
*/
class CssToAttributeConverter extends AbstractHtmlProcessor
{
/**
* This multi-level array contains simple mappings of CSS properties to
* HTML attributes. If a mapping only applies to certain HTML nodes or
* only for certain values, the mapping is an object with a whitelist
* of nodes and values.
*
* @var array<string, array{attribute: string, nodes?: array<int, string>, values?: array<int, string>}>
*/
private $cssToHtmlMap = [
'background-color' => [
'attribute' => 'bgcolor',
],
'text-align' => [
'attribute' => 'align',
'nodes' => ['p', 'div', 'td', 'th'],
'values' => ['left', 'right', 'center', 'justify'],
],
'float' => [
'attribute' => 'align',
'nodes' => ['table', 'img'],
'values' => ['left', 'right'],
],
'border-spacing' => [
'attribute' => 'cellspacing',
'nodes' => ['table'],
],
];
/**
* @var array<string, array<string, string>>
*/
private static $parsedCssCache = [];
/**
* Maps the CSS from the style nodes to visual HTML attributes.
*
* @return self fluent interface
*/
public function convertCssToVisualAttributes(): self
{
/** @var \DOMElement $node */
foreach ($this->getAllNodesWithStyleAttribute() as $node) {
$inlineStyleDeclarations = $this->parseCssDeclarationsBlock($node->getAttribute('style'));
$this->mapCssToHtmlAttributes($inlineStyleDeclarations, $node);
}
return $this;
}
/**
* Returns a list with all DOM nodes that have a style attribute.
*
* @return \DOMNodeList
*/
private function getAllNodesWithStyleAttribute(): \DOMNodeList
{
return $this->getXPath()->query('//*[@style]');
}
/**
* Parses a CSS declaration block into property name/value pairs.
*
* Example:
*
* The declaration block
*
* "color: #000; font-weight: bold;"
*
* will be parsed into the following array:
*
* "color" => "#000"
* "font-weight" => "bold"
*
* @param string $cssDeclarationsBlock the CSS declarations block without the curly braces, may be empty
*
* @return array<string, string>
* the CSS declarations with the property names as array keys and the property values as array values
*/
private function parseCssDeclarationsBlock(string $cssDeclarationsBlock): array
{
if (isset(self::$parsedCssCache[$cssDeclarationsBlock])) {
return self::$parsedCssCache[$cssDeclarationsBlock];
}
$properties = [];
foreach (\preg_split('/;(?!base64|charset)/', $cssDeclarationsBlock) as $declaration) {
/** @var array<int, string> $matches */
$matches = [];
if (!\preg_match('/^([A-Za-z\\-]+)\\s*:\\s*(.+)$/s', \trim($declaration), $matches)) {
continue;
}
$propertyName = \strtolower($matches[1]);
$propertyValue = $matches[2];
$properties[$propertyName] = $propertyValue;
}
self::$parsedCssCache[$cssDeclarationsBlock] = $properties;
return $properties;
}
/**
* Applies $styles to $node.
*
* This method maps CSS styles to HTML attributes and adds those to the
* node.
*
* @param array<string, string> $styles the new CSS styles taken from the global styles to be applied to this node
* @param \DOMElement $node node to apply styles to
*/
private function mapCssToHtmlAttributes(array $styles, \DOMElement $node): void
{
foreach ($styles as $property => $value) {
// Strip !important indicator
$value = \trim(\str_replace('!important', '', $value));
$this->mapCssToHtmlAttribute($property, $value, $node);
}
}
/**
* Tries to apply the CSS style to $node as an attribute.
*
* This method maps a CSS rule to HTML attributes and adds those to the node.
*
* @param string $property the name of the CSS property to map
* @param string $value the value of the style rule to map
* @param \DOMElement $node node to apply styles to
*/
private function mapCssToHtmlAttribute(string $property, string $value, \DOMElement $node): void
{
if (!$this->mapSimpleCssProperty($property, $value, $node)) {
$this->mapComplexCssProperty($property, $value, $node);
}
}
/**
* Looks up the CSS property in the mapping table and maps it if it matches the conditions.
*
* @param string $property the name of the CSS property to map
* @param string $value the value of the style rule to map
* @param \DOMElement $node node to apply styles to
*
* @return bool true if the property can be mapped using the simple mapping table
*/
private function mapSimpleCssProperty(string $property, string $value, \DOMElement $node): bool
{
if (!isset($this->cssToHtmlMap[$property])) {
return false;
}
$mapping = $this->cssToHtmlMap[$property];
$nodesMatch = !isset($mapping['nodes']) || \in_array($node->nodeName, $mapping['nodes'], true);
$valuesMatch = !isset($mapping['values']) || \in_array($value, $mapping['values'], true);
$canBeMapped = $nodesMatch && $valuesMatch;
if ($canBeMapped) {
$node->setAttribute($mapping['attribute'], $value);
}
return $canBeMapped;
}
/**
* Maps CSS properties that need special transformation to an HTML attribute.
*
* @param string $property the name of the CSS property to map
* @param string $value the value of the style rule to map
* @param \DOMElement $node node to apply styles to
*/
private function mapComplexCssProperty(string $property, string $value, \DOMElement $node): void
{
switch ($property) {
case 'background':
$this->mapBackgroundProperty($node, $value);
break;
case 'width':
// intentional fall-through
case 'height':
$this->mapWidthOrHeightProperty($node, $value, $property);
break;
case 'margin':
$this->mapMarginProperty($node, $value);
break;
case 'border':
$this->mapBorderProperty($node, $value);
break;
default:
}
}
/**
* @param \DOMElement $node node to apply styles to
* @param string $value the value of the style rule to map
*/
private function mapBackgroundProperty(\DOMElement $node, string $value): void
{
// parse out the color, if any
/** @var array<int, string> $styles */
$styles = \explode(' ', $value, 2);
$first = $styles[0];
if (\is_numeric($first[0]) || \strncmp($first, 'url', 3) === 0) {
return;
}
// as this is not a position or image, assume it's a color
$node->setAttribute('bgcolor', $first);
}
/**
* @param \DOMElement $node node to apply styles to
* @param string $value the value of the style rule to map
* @param string $property the name of the CSS property to map
*/
private function mapWidthOrHeightProperty(\DOMElement $node, string $value, string $property): void
{
// only parse values in px and %, but not values like "auto"
if (!\preg_match('/^(\\d+)(\\.(\\d+))?(px|%)$/', $value)) {
return;
}
$number = \preg_replace('/[^0-9.%]/', '', $value);
$node->setAttribute($property, $number);
}
/**
* @param \DOMElement $node node to apply styles to
* @param string $value the value of the style rule to map
*/
private function mapMarginProperty(\DOMElement $node, string $value): void
{
if (!$this->isTableOrImageNode($node)) {
return;
}
$margins = $this->parseCssShorthandValue($value);
if ($margins['left'] === 'auto' && $margins['right'] === 'auto') {
$node->setAttribute('align', 'center');
}
}
/**
* @param \DOMElement $node node to apply styles to
* @param string $value the value of the style rule to map
*/
private function mapBorderProperty(\DOMElement $node, string $value): void
{
if (!$this->isTableOrImageNode($node)) {
return;
}
if ($value === 'none' || $value === '0') {
$node->setAttribute('border', '0');
}
}
/**
* @param \DOMElement $node
*
* @return bool
*/
private function isTableOrImageNode(\DOMElement $node): bool
{
return $node->nodeName === 'table' || $node->nodeName === 'img';
}
/**
* Parses a shorthand CSS value and splits it into individual values. For example: `padding: 0 auto;` - `0 auto` is
* split into top: 0, left: auto, bottom: 0, right: auto.
*
* @param string $value a CSS property value with 1, 2, 3 or 4 sizes
*
* @return array<string, string>
* an array of values for top, right, bottom and left (using these as associative array keys)
*/
private function parseCssShorthandValue(string $value): array
{
/** @var array<int, string> $values */
$values = \preg_split('/\\s+/', $value);
$css = [];
$css['top'] = $values[0];
$css['right'] = (\count($values) > 1) ? $values[1] : $css['top'];
$css['bottom'] = (\count($values) > 2) ? $values[2] : $css['top'];
$css['left'] = (\count($values) > 3) ? $values[3] : $css['right'];
return $css;
}
}
HtmlNormalizer.php 0000644 00000000502 15155373351 0010231 0 ustar 00 <?php
declare(strict_types=1);
namespace Pelago\Emogrifier\HtmlProcessor;
/**
* Normalizes HTML:
* - add a document type (HTML5) if missing
* - disentangle incorrectly nested tags
* - add HEAD and BODY elements (if they are missing)
* - reformat the HTML
*/
class HtmlNormalizer extends AbstractHtmlProcessor
{
}
HtmlPruner.php 0000644 00000012010 15155373351 0007357 0 ustar 00 <?php
declare(strict_types=1);
namespace Pelago\Emogrifier\HtmlProcessor;
use Pelago\Emogrifier\CssInliner;
use Pelago\Emogrifier\Utilities\ArrayIntersector;
/**
* This class can remove things from HTML.
*/
class HtmlPruner extends AbstractHtmlProcessor
{
/**
* We need to look for display:none, but we need to do a case-insensitive search. Since DOMDocument only
* supports XPath 1.0, lower-case() isn't available to us. We've thus far only set attributes to lowercase,
* not attribute values. Consequently, we need to translate() the letters that would be in 'NONE' ("NOE")
* to lowercase.
*
* @var string
*/
private const DISPLAY_NONE_MATCHER
= '//*[@style and contains(translate(translate(@style," ",""),"NOE","noe"),"display:none")'
. ' and not(@class and contains(concat(" ", normalize-space(@class), " "), " -emogrifier-keep "))]';
/**
* Removes elements that have a "display: none;" style.
*
* @return self fluent interface
*/
public function removeElementsWithDisplayNone(): self
{
$elementsWithStyleDisplayNone = $this->getXPath()->query(self::DISPLAY_NONE_MATCHER);
if ($elementsWithStyleDisplayNone->length === 0) {
return $this;
}
foreach ($elementsWithStyleDisplayNone as $element) {
$parentNode = $element->parentNode;
if ($parentNode !== null) {
$parentNode->removeChild($element);
}
}
return $this;
}
/**
* Removes classes that are no longer required (e.g. because there are no longer any CSS rules that reference them)
* from `class` attributes.
*
* Note that this does not inspect the CSS, but expects to be provided with a list of classes that are still in use.
*
* This method also has the (presumably beneficial) side-effect of minifying (removing superfluous whitespace from)
* `class` attributes.
*
* @param array<array-key, string> $classesToKeep names of classes that should not be removed
*
* @return self fluent interface
*/
public function removeRedundantClasses(array $classesToKeep = []): self
{
$elementsWithClassAttribute = $this->getXPath()->query('//*[@class]');
if ($classesToKeep !== []) {
$this->removeClassesFromElements($elementsWithClassAttribute, $classesToKeep);
} else {
// Avoid unnecessary processing if there are no classes to keep.
$this->removeClassAttributeFromElements($elementsWithClassAttribute);
}
return $this;
}
/**
* Removes classes from the `class` attribute of each element in `$elements`, except any in `$classesToKeep`,
* removing the `class` attribute itself if the resultant list is empty.
*
* @param \DOMNodeList $elements
* @param array<array-key, string> $classesToKeep
*/
private function removeClassesFromElements(\DOMNodeList $elements, array $classesToKeep): void
{
$classesToKeepIntersector = new ArrayIntersector($classesToKeep);
/** @var \DOMElement $element */
foreach ($elements as $element) {
$elementClasses = \preg_split('/\\s++/', \trim($element->getAttribute('class')));
$elementClassesToKeep = $classesToKeepIntersector->intersectWith($elementClasses);
if ($elementClassesToKeep !== []) {
$element->setAttribute('class', \implode(' ', $elementClassesToKeep));
} else {
$element->removeAttribute('class');
}
}
}
/**
* Removes the `class` attribute from each element in `$elements`.
*
* @param \DOMNodeList $elements
*/
private function removeClassAttributeFromElements(\DOMNodeList $elements): void
{
/** @var \DOMElement $element */
foreach ($elements as $element) {
$element->removeAttribute('class');
}
}
/**
* After CSS has been inlined, there will likely be some classes in `class` attributes that are no longer referenced
* by any remaining (uninlinable) CSS. This method removes such classes.
*
* Note that it does not inspect the remaining CSS, but uses information readily available from the `CssInliner`
* instance about the CSS rules that could not be inlined.
*
* @param CssInliner $cssInliner object instance that performed the CSS inlining
*
* @return self fluent interface
*
* @throws \BadMethodCallException if `inlineCss` has not first been called on `$cssInliner`
*/
public function removeRedundantClassesAfterCssInlined(CssInliner $cssInliner): self
{
$classesToKeepAsKeys = [];
foreach ($cssInliner->getMatchingUninlinableSelectors() as $selector) {
\preg_match_all('/\\.(-?+[_a-zA-Z][\\w\\-]*+)/', $selector, $matches);
$classesToKeepAsKeys += \array_fill_keys($matches[1], true);
}
$this->removeRedundantClasses(\array_keys($classesToKeepAsKeys));
return $this;
}
}