CssInliner.php — WooCommerce 7.4.0-rc.2

woocommerce / vendor / pelago / emogrifier / src / CssInliner.php

woocommerce / vendor / pelago / emogrifier / src Last commit date

Caching 4 years ago Css 4 years ago HtmlProcessor 4 years ago Utilities 4 years ago CssInliner.php 4 years ago

CssInliner.php

1197 lines

1	<?php
2
3	declare(strict_types=1);
4
5	namespace Pelago\Emogrifier;
6
7	use Pelago\Emogrifier\Css\CssDocument;
8	use Pelago\Emogrifier\HtmlProcessor\AbstractHtmlProcessor;
9	use Pelago\Emogrifier\Utilities\CssConcatenator;
10	use Symfony\Component\CssSelector\CssSelectorConverter;
11	use Symfony\Component\CssSelector\Exception\ParseException;
12
13	/**
14	* This class provides functions for converting CSS styles into inline style attributes in your HTML code.
15	*/
16	class CssInliner extends AbstractHtmlProcessor
17	{
18	/**
19	* @var int
20	*/
21	private const CACHE_KEY_SELECTOR = 0;
22
23	/**
24	* @var int
25	*/
26	private const CACHE_KEY_CSS_DECLARATIONS_BLOCK = 1;
27
28	/**
29	* @var int
30	*/
31	private const CACHE_KEY_COMBINED_STYLES = 2;
32
33	/**
34	* Regular expression component matching a static pseudo class in a selector, without the preceding ":",
35	* for which the applicable elements can be determined (by converting the selector to an XPath expression).
36	* (Contains alternation without a group and is intended to be placed within a capturing, non-capturing or lookahead
37	* group, as appropriate for the usage context.)
38	*
39	* @var string
40	*/
41	private const PSEUDO_CLASS_MATCHER
42	= 'empty\|(?:first\|last\|nth(?:-last)?+\|only)-(?:child\|of-type)\|not\$[[:ascii:]]*\$';
43
44	/**
45	* This regular expression componenet matches an `...of-type` pseudo class name, without the preceding ":". These
46	* pseudo-classes can currently online be inlined if they have an associated type in the selector expression.
47	*
48	* @var string
49	*/
50	private const OF_TYPE_PSEUDO_CLASS_MATCHER = '(?:first\|last\|nth(?:-last)?+\|only)-of-type';
51
52	/**
53	* regular expression component to match a selector combinator
54	*
55	* @var string
56	*/
57	private const COMBINATOR_MATCHER = '(?:\\s++\|\\s+[>+~]\\s+)(?=[[:alpha:]_\\-.#*:\\[])';
58
59	/**
60	* @var array<string, bool>
61	*/
62	private $excludedSelectors = [];
63
64	/**
65	* @var array<string, bool>
66	*/
67	private $allowedMediaTypes = ['all' => true, 'screen' => true, 'print' => true];
68
69	/**
70	* @var array{
71	* 0: array<string, int>,
72	* 1: array<string, array<string, string>>,
73	* 2: array<string, string>
74	* }
75	*/
76	private $caches = [
77	self::CACHE_KEY_SELECTOR => [],
78	self::CACHE_KEY_CSS_DECLARATIONS_BLOCK => [],
79	self::CACHE_KEY_COMBINED_STYLES => [],
80	];
81
82	/**
83	* @var ?CssSelectorConverter
84	*/
85	private $cssSelectorConverter = null;
86
87	/**
88	* the visited nodes with the XPath paths as array keys
89	*
90	* @var array<string, \DOMElement>
91	*/
92	private $visitedNodes = [];
93
94	/**
95	* the styles to apply to the nodes with the XPath paths as array keys for the outer array
96	* and the attribute names/values as key/value pairs for the inner array
97	*
98	* @var array<string, array<string, string>>
99	*/
100	private $styleAttributesForNodes = [];
101
102	/**
103	* Determines whether the "style" attributes of tags in the the HTML passed to this class should be preserved.
104	* If set to false, the value of the style attributes will be discarded.
105	*
106	* @var bool
107	*/
108	private $isInlineStyleAttributesParsingEnabled = true;
109
110	/**
111	* Determines whether the `<style>` blocks in the HTML passed to this class should be parsed.
112	*
113	* If set to true, the `<style>` blocks will be removed from the HTML and their contents will be applied to the HTML
114	* via inline styles.
115	*
116	* If set to false, the `<style>` blocks will be left as they are in the HTML.
117	*
118	* @var bool
119	*/
120	private $isStyleBlocksParsingEnabled = true;
121
122	/**
123	* For calculating selector precedence order.
124	* Keys are a regular expression part to match before a CSS name.
125	* Values are a multiplier factor per match to weight specificity.
126	*
127	* @var array<string, int>
128	*/
129	private $selectorPrecedenceMatchers = [
130	// IDs: worth 10000
131	'\\#' => 10000,
132	// classes, attributes, pseudo-classes (not pseudo-elements) except `:not`: worth 100
133	'(?:\\.\|\\[\|(?<!:):(?!not\\())' => 100,
134	// elements (not attribute values or `:not`), pseudo-elements: worth 1
135	'(?:(?<![="\':\\w\\-])\|::)' => 1,
136	];
137
138	/**
139	* array of data describing CSS rules which apply to the document but cannot be inlined, in the format returned by
140	* {@see collateCssRules}
141	*
142	* @var array<array-key, array{
143	* media: string,
144	* selector: string,
145	* hasUnmatchablePseudo: bool,
146	* declarationsBlock: string,
147	* line: int
148	* }>\|null
149	*/
150	private $matchingUninlinableCssRules = null;
151
152	/**
153	* Emogrifier will throw Exceptions when it encounters an error instead of silently ignoring them.
154	*
155	* @var bool
156	*/
157	private $debug = false;
158
159	/**
160	* Inlines the given CSS into the existing HTML.
161	*
162	* @param string $css the CSS to inline, must be UTF-8-encoded
163	*
164	* @return self fluent interface
165	*
166	* @throws ParseException in debug mode, if an invalid selector is encountered
167	* @throws \RuntimeException in debug mode, if an internal PCRE error occurs
168	*/
169	public function inlineCss(string $css = ''): self
170	{
171	$this->clearAllCaches();
172	$this->purgeVisitedNodes();
173
174	$this->normalizeStyleAttributesOfAllNodes();
175
176	$combinedCss = $css;
177	// grab any existing style blocks from the HTML and append them to the existing CSS
178	// (these blocks should be appended so as to have precedence over conflicting styles in the existing CSS)
179	if ($this->isStyleBlocksParsingEnabled) {
180	$combinedCss .= $this->getCssFromAllStyleNodes();
181	}
182	$parsedCss = new CssDocument($combinedCss);
183
184	$excludedNodes = $this->getNodesToExclude();
185	$cssRules = $this->collateCssRules($parsedCss);
186	$cssSelectorConverter = $this->getCssSelectorConverter();
187	foreach ($cssRules['inlinable'] as $cssRule) {
188	try {
189	$nodesMatchingCssSelectors = $this->getXPath()
190	->query($cssSelectorConverter->toXPath($cssRule['selector']));
191
192	/** @var \DOMElement $node */
193	foreach ($nodesMatchingCssSelectors as $node) {
194	if (\in_array($node, $excludedNodes, true)) {
195	continue;
196	}
197	$this->copyInlinableCssToStyleAttribute($node, $cssRule);
198	}
199	} catch (ParseException $e) {
200	if ($this->debug) {
201	throw $e;
202	}
203	}
204	}
205
206	if ($this->isInlineStyleAttributesParsingEnabled) {
207	$this->fillStyleAttributesWithMergedStyles();
208	}
209
210	$this->removeImportantAnnotationFromAllInlineStyles();
211
212	$this->determineMatchingUninlinableCssRules($cssRules['uninlinable']);
213	$this->copyUninlinableCssToStyleNode($parsedCss);
214
215	return $this;
216	}
217
218	/**
219	* Disables the parsing of inline styles.
220	*
221	* @return self fluent interface
222	*/
223	public function disableInlineStyleAttributesParsing(): self
224	{
225	$this->isInlineStyleAttributesParsingEnabled = false;
226
227	return $this;
228	}
229
230	/**
231	* Disables the parsing of `<style>` blocks.
232	*
233	* @return self fluent interface
234	*/
235	public function disableStyleBlocksParsing(): self
236	{
237	$this->isStyleBlocksParsingEnabled = false;
238
239	return $this;
240	}
241
242	/**
243	* Marks a media query type to keep.
244	*
245	* @param string $mediaName the media type name, e.g., "braille"
246	*
247	* @return self fluent interface
248	*/
249	public function addAllowedMediaType(string $mediaName): self
250	{
251	$this->allowedMediaTypes[$mediaName] = true;
252
253	return $this;
254	}
255
256	/**
257	* Drops a media query type from the allowed list.
258	*
259	* @param string $mediaName the tag name, e.g., "braille"
260	*
261	* @return self fluent interface
262	*/
263	public function removeAllowedMediaType(string $mediaName): self
264	{
265	if (isset($this->allowedMediaTypes[$mediaName])) {
266	unset($this->allowedMediaTypes[$mediaName]);
267	}
268
269	return $this;
270	}
271
272	/**
273	* Adds a selector to exclude nodes from emogrification.
274	*
275	* Any nodes that match the selector will not have their style altered.
276	*
277	* @param string $selector the selector to exclude, e.g., ".editor"
278	*
279	* @return self fluent interface
280	*/
281	public function addExcludedSelector(string $selector): self
282	{
283	$this->excludedSelectors[$selector] = true;
284
285	return $this;
286	}
287
288	/**
289	* No longer excludes the nodes matching this selector from emogrification.
290	*
291	* @param string $selector the selector to no longer exclude, e.g., ".editor"
292	*
293	* @return self fluent interface
294	*/
295	public function removeExcludedSelector(string $selector): self
296	{
297	if (isset($this->excludedSelectors[$selector])) {
298	unset($this->excludedSelectors[$selector]);
299	}
300
301	return $this;
302	}
303
304	/**
305	* Sets the debug mode.
306	*
307	* @param bool $debug set to true to enable debug mode
308	*
309	* @return self fluent interface
310	*/
311	public function setDebug(bool $debug): self
312	{
313	$this->debug = $debug;
314
315	return $this;
316	}
317
318	/**
319	* Gets the array of selectors present in the CSS provided to `inlineCss()` for which the declarations could not be
320	* applied as inline styles, but which may affect elements in the HTML. The relevant CSS will have been placed in a
321	* `<style>` element. The selectors may include those used within `@media` rules or those involving dynamic
322	* pseudo-classes (such as `:hover`) or pseudo-elements (such as `::after`).
323	*
324	* @return array<array-key, string>
325	*
326	* @throws \BadMethodCallException if `inlineCss` has not been called first
327	*/
328	public function getMatchingUninlinableSelectors(): array
329	{
330	return \array_column($this->getMatchingUninlinableCssRules(), 'selector');
331	}
332
333	/**
334	* @return array<array-key, array{
335	* media: string,
336	* selector: string,
337	* hasUnmatchablePseudo: bool,
338	* declarationsBlock: string,
339	* line: int
340	* }>
341	*
342	* @throws \BadMethodCallException if `inlineCss` has not been called first
343	*/
344	private function getMatchingUninlinableCssRules(): array
345	{
346	if (!\is_array($this->matchingUninlinableCssRules)) {
347	throw new \BadMethodCallException('inlineCss must be called first', 1568385221);
348	}
349
350	return $this->matchingUninlinableCssRules;
351	}
352
353	/**
354	* Clears all caches.
355	*/
356	private function clearAllCaches(): void
357	{
358	$this->caches = [
359	self::CACHE_KEY_SELECTOR => [],
360	self::CACHE_KEY_CSS_DECLARATIONS_BLOCK => [],
361	self::CACHE_KEY_COMBINED_STYLES => [],
362	];
363	}
364
365	/**
366	* Purges the visited nodes.
367	*/
368	private function purgeVisitedNodes(): void
369	{
370	$this->visitedNodes = [];
371	$this->styleAttributesForNodes = [];
372	}
373
374	/**
375	* Parses the document and normalizes all existing CSS attributes.
376	* This changes 'DISPLAY: none' to 'display: none'.
377	* We wouldn't have to do this if DOMXPath supported XPath 2.0.
378	* Also stores a reference of nodes with existing inline styles so we don't overwrite them.
379	*/
380	private function normalizeStyleAttributesOfAllNodes(): void
381	{
382	/** @var \DOMElement $node */
383	foreach ($this->getAllNodesWithStyleAttribute() as $node) {
384	if ($this->isInlineStyleAttributesParsingEnabled) {
385	$this->normalizeStyleAttributes($node);
386	}
387	// Remove style attribute in every case, so we can add them back (if inline style attributes
388	// parsing is enabled) to the end of the style list, thus keeping the right priority of CSS rules;
389	// else original inline style rules may remain at the beginning of the final inline style definition
390	// of a node, which may give not the desired results
391	$node->removeAttribute('style');
392	}
393	}
394
395	/**
396	* Returns a list with all DOM nodes that have a style attribute.
397	*
398	* @return \DOMNodeList
399	*
400	* @throws \RuntimeException
401	*/
402	private function getAllNodesWithStyleAttribute(): \DOMNodeList
403	{
404	$query = '//*[@style]';
405	$matches = $this->getXPath()->query($query);
406	if (!$matches instanceof \DOMNodeList) {
407	throw new \RuntimeException('XPatch query failed: ' . $query, 1618577797);
408	}
409
410	return $matches;
411	}
412
413	/**
414	* Normalizes the value of the "style" attribute and saves it.
415	*
416	* @param \DOMElement $node
417	*/
418	private function normalizeStyleAttributes(\DOMElement $node): void
419	{
420	$normalizedOriginalStyle = \preg_replace_callback(
421	'/-?+[_a-zA-Z][\\w\\-]*+(?=:)/S',
422	/** @param array<array-key, string> $propertyNameMatches */
423	static function (array $propertyNameMatches): string {
424	return \strtolower($propertyNameMatches[0]);
425	},
426	$node->getAttribute('style')
427	);
428
429	// In order to not overwrite existing style attributes in the HTML, we have to save the original HTML styles.
430	$nodePath = $node->getNodePath();
431	if (\is_string($nodePath) && !isset($this->styleAttributesForNodes[$nodePath])) {
432	$this->styleAttributesForNodes[$nodePath] = $this->parseCssDeclarationsBlock($normalizedOriginalStyle);
433	$this->visitedNodes[$nodePath] = $node;
434	}
435
436	$node->setAttribute('style', $normalizedOriginalStyle);
437	}
438
439	/**
440	* Parses a CSS declaration block into property name/value pairs.
441	*
442	* Example:
443	*
444	* The declaration block
445	*
446	* "color: #000; font-weight: bold;"
447	*
448	* will be parsed into the following array:
449	*
450	* "color" => "#000"
451	* "font-weight" => "bold"
452	*
453	* @param string $cssDeclarationsBlock the CSS declarations block without the curly braces, may be empty
454	*
455	* @return array<string, string>
456	* the CSS declarations with the property names as array keys and the property values as array values
457	*/
458	private function parseCssDeclarationsBlock(string $cssDeclarationsBlock): array
459	{
460	if (isset($this->caches[self::CACHE_KEY_CSS_DECLARATIONS_BLOCK][$cssDeclarationsBlock])) {
461	return $this->caches[self::CACHE_KEY_CSS_DECLARATIONS_BLOCK][$cssDeclarationsBlock];
462	}
463
464	$properties = [];
465	foreach (\preg_split('/;(?!base64\|charset)/', $cssDeclarationsBlock) as $declaration) {
466	/** @var array<int, string> $matches */
467	$matches = [];
468	if (!\preg_match('/^([A-Za-z\\-]+)\\s:\\s(.+)$/s', \trim($declaration), $matches)) {
469	continue;
470	}
471
472	$propertyName = \strtolower($matches[1]);
473	$propertyValue = $matches[2];
474	$properties[$propertyName] = $propertyValue;
475	}
476	$this->caches[self::CACHE_KEY_CSS_DECLARATIONS_BLOCK][$cssDeclarationsBlock] = $properties;
477
478	return $properties;
479	}
480
481	/**
482	* Returns CSS content.
483	*
484	* @return string
485	*/
486	private function getCssFromAllStyleNodes(): string
487	{
488	$styleNodes = $this->getXPath()->query('//style');
489	if ($styleNodes === false) {
490	return '';
491	}
492
493	$css = '';
494	foreach ($styleNodes as $styleNode) {
495	$css .= "\n\n" . $styleNode->nodeValue;
496	$parentNode = $styleNode->parentNode;
497	if ($parentNode instanceof \DOMNode) {
498	$parentNode->removeChild($styleNode);
499	}
500	}
501
502	return $css;
503	}
504
505	/**
506	* Find the nodes that are not to be emogrified.
507	*
508	* @return array<int, \DOMElement>
509	*
510	* @throws ParseException
511	* @throws \UnexpectedValueException
512	*/
513	private function getNodesToExclude(): array
514	{
515	$excludedNodes = [];
516	foreach (\array_keys($this->excludedSelectors) as $selectorToExclude) {
517	try {
518	$matchingNodes = $this->getXPath()
519	->query($this->getCssSelectorConverter()->toXPath($selectorToExclude));
520
521	foreach ($matchingNodes as $node) {
522	if (!$node instanceof \DOMElement) {
523	$path = $node->getNodePath() ?? '$node';
524	throw new \UnexpectedValueException($path . ' is not a DOMElement.', 1617975914);
525	}
526	$excludedNodes[] = $node;
527	}
528	} catch (ParseException $e) {
529	if ($this->debug) {
530	throw $e;
531	}
532	}
533	}
534
535	return $excludedNodes;
536	}
537
538	/**
539	* @return CssSelectorConverter
540	*/
541	private function getCssSelectorConverter(): CssSelectorConverter
542	{
543	if (!$this->cssSelectorConverter instanceof CssSelectorConverter) {
544	$this->cssSelectorConverter = new CssSelectorConverter();
545	}
546
547	return $this->cssSelectorConverter;
548	}
549
550	/**
551	* Collates the individual rules from a `CssDocument` object.
552	*
553	* @param CssDocument $parsedCss
554	*
555	* @return array<string, array<array-key, array{
556	* media: string,
557	* selector: string,
558	* hasUnmatchablePseudo: bool,
559	* declarationsBlock: string,
560	* line: int
561	* }>>
562	* This 2-entry array has the key "inlinable" containing rules which can be inlined as `style` attributes
563	* and the key "uninlinable" containing rules which cannot. Each value is an array of sub-arrays with the
564	* following keys:
565	* - "media" (the media query string, e.g. "@media screen and (max-width: 480px)",
566	* or an empty string if not from a `@media` rule);
567	* - "selector" (the CSS selector, e.g., "*" or "header h1");
568	* - "hasUnmatchablePseudo" (`true` if that selector contains pseudo-elements or dynamic pseudo-classes such
569	* that the declarations cannot be applied inline);
570	* - "declarationsBlock" (the semicolon-separated CSS declarations for that selector,
571	* e.g., `color: red; height: 4px;`);
572	* - "line" (the line number, e.g. 42).
573	*/
574	private function collateCssRules(CssDocument $parsedCss): array
575	{
576	$matches = $parsedCss->getStyleRulesData(\array_keys($this->allowedMediaTypes));
577
578	$cssRules = [
579	'inlinable' => [],
580	'uninlinable' => [],
581	];
582	foreach ($matches as $key => $cssRule) {
583	if (!$cssRule->hasAtLeastOneDeclaration()) {
584	continue;
585	}
586
587	$mediaQuery = $cssRule->getContainingAtRule();
588	$declarationsBlock = $cssRule->getDeclarationAsText();
589	foreach ($cssRule->getSelectors() 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	$hasUnmatchablePseudo = $hasPseudoElement \|\| $this->hasUnsupportedPseudoClass($selector);
594
595	$parsedCssRule = [
596	'media' => $mediaQuery,
597	'selector' => $selector,
598	'hasUnmatchablePseudo' => $hasUnmatchablePseudo,
599	'declarationsBlock' => $declarationsBlock,
600	// keep track of where it appears in the file, since order is important
601	'line' => $key,
602	];
603	$ruleType = (!$cssRule->hasContainingAtRule() && !$hasUnmatchablePseudo) ? 'inlinable' : 'uninlinable';
604	$cssRules[$ruleType][] = $parsedCssRule;
605	}
606	}
607
608	\usort(
609	$cssRules['inlinable'],
610	/**
611	* @param array{selector: string, line: int} $first
612	* @param array{selector: string, line: int} $second
613	*/
614	function (array $first, array $second): int {
615	return $this->sortBySelectorPrecedence($first, $second);
616	}
617	);
618
619	return $cssRules;
620	}
621
622	/**
623	* Tests if a selector contains a pseudo-class which would mean it cannot be converted to an XPath expression for
624	* inlining CSS declarations.
625	*
626	* Any pseudo class that does not match {@see PSEUDO_CLASS_MATCHER} cannot be converted. Additionally, `...of-type`
627	* pseudo-classes cannot be converted if they are not associated with a type selector.
628	*
629	* @param string $selector
630	*
631	* @return bool
632	*/
633	private function hasUnsupportedPseudoClass(string $selector): bool
634	{
635	if (\preg_match('/:(?!' . self::PSEUDO_CLASS_MATCHER . ')[\\w\\-]/i', $selector)) {
636	return true;
637	}
638
639	if (!\preg_match('/:(?:' . self::OF_TYPE_PSEUDO_CLASS_MATCHER . ')/i', $selector)) {
640	return false;
641	}
642
643	foreach (\preg_split('/' . self::COMBINATOR_MATCHER . '/', $selector) as $selectorPart) {
644	if ($this->selectorPartHasUnsupportedOfTypePseudoClass($selectorPart)) {
645	return true;
646	}
647	}
648
649	return false;
650	}
651
652	/**
653	* Tests if part of a selector contains an `...of-type` pseudo-class such that it cannot be converted to an XPath
654	* expression.
655	*
656	* @param string $selectorPart part of a selector which has been split up at combinators
657	*
658	* @return bool `true` if the selector part does not have a type but does have an `...of-type` pseudo-class
659	*/
660	private function selectorPartHasUnsupportedOfTypePseudoClass(string $selectorPart): bool
661	{
662	if (\preg_match('/^[\\w\\-]/', $selectorPart)) {
663	return false;
664	}
665
666	return (bool)\preg_match('/:(?:' . self::OF_TYPE_PSEUDO_CLASS_MATCHER . ')/i', $selectorPart);
667	}
668
669	/**
670	* @param array{selector: string, line: int} $first
671	* @param array{selector: string, line: int} $second
672	*
673	* @return int
674	*/
675	private function sortBySelectorPrecedence(array $first, array $second): int
676	{
677	$precedenceOfFirst = $this->getCssSelectorPrecedence($first['selector']);
678	$precedenceOfSecond = $this->getCssSelectorPrecedence($second['selector']);
679
680	// We want these sorted in ascending order so selectors with lesser precedence get processed first and
681	// selectors with greater precedence get sorted last.
682	$precedenceForEquals = $first['line'] < $second['line'] ? -1 : 1;
683	$precedenceForNotEquals = $precedenceOfFirst < $precedenceOfSecond ? -1 : 1;
684	return ($precedenceOfFirst === $precedenceOfSecond) ? $precedenceForEquals : $precedenceForNotEquals;
685	}
686
687	/**
688	* @param string $selector
689	*
690	* @return int
691	*/
692	private function getCssSelectorPrecedence(string $selector): int
693	{
694	$selectorKey = \md5($selector);
695	if (isset($this->caches[self::CACHE_KEY_SELECTOR][$selectorKey])) {
696	return $this->caches[self::CACHE_KEY_SELECTOR][$selectorKey];
697	}
698
699	$precedence = 0;
700	foreach ($this->selectorPrecedenceMatchers as $matcher => $value) {
701	if (\trim($selector) === '') {
702	break;
703	}
704	$number = 0;
705	$selector = \preg_replace('/' . $matcher . '\\w+/', '', $selector, -1, $number);
706	$precedence += ($value * (int)$number);
707	}
708	$this->caches[self::CACHE_KEY_SELECTOR][$selectorKey] = $precedence;
709
710	return $precedence;
711	}
712
713	/**
714	* Copies $cssRule into the style attribute of $node.
715	*
716	* Note: This method does not check whether $cssRule matches $node.
717	*
718	* @param \DOMElement $node
719	* @param array{
720	* media: string,
721	* selector: string,
722	* hasUnmatchablePseudo: bool,
723	* declarationsBlock: string,
724	* line: int
725	* } $cssRule
726	*/
727	private function copyInlinableCssToStyleAttribute(\DOMElement $node, array $cssRule): void
728	{
729	$declarationsBlock = $cssRule['declarationsBlock'];
730	$newStyleDeclarations = $this->parseCssDeclarationsBlock($declarationsBlock);
731	if ($newStyleDeclarations === []) {
732	return;
733	}
734
735	// if it has a style attribute, get it, process it, and append (overwrite) new stuff
736	if ($node->hasAttribute('style')) {
737	// break it up into an associative array
738	$oldStyleDeclarations = $this->parseCssDeclarationsBlock($node->getAttribute('style'));
739	} else {
740	$oldStyleDeclarations = [];
741	}
742	$node->setAttribute(
743	'style',
744	$this->generateStyleStringFromDeclarationsArrays($oldStyleDeclarations, $newStyleDeclarations)
745	);
746	}
747
748	/**
749	* This method merges old or existing name/value array with new name/value array
750	* and then generates a string of the combined style suitable for placing inline.
751	* This becomes the single point for CSS string generation allowing for consistent
752	* CSS output no matter where the CSS originally came from.
753	*
754	* @param array<string, string> $oldStyles
755	* @param array<string, string> $newStyles
756	*
757	* @return string
758	*/
759	private function generateStyleStringFromDeclarationsArrays(array $oldStyles, array $newStyles): string
760	{
761	$cacheKey = \serialize([$oldStyles, $newStyles]);
762	if (isset($this->caches[self::CACHE_KEY_COMBINED_STYLES][$cacheKey])) {
763	return $this->caches[self::CACHE_KEY_COMBINED_STYLES][$cacheKey];
764	}
765
766	// Unset the overridden styles to preserve order, important if shorthand and individual properties are mixed
767	foreach ($oldStyles as $attributeName => $attributeValue) {
768	if (!isset($newStyles[$attributeName])) {
769	continue;
770	}
771
772	$newAttributeValue = $newStyles[$attributeName];
773	if (
774	$this->attributeValueIsImportant($attributeValue)
775	&& !$this->attributeValueIsImportant($newAttributeValue)
776	) {
777	unset($newStyles[$attributeName]);
778	} else {
779	unset($oldStyles[$attributeName]);
780	}
781	}
782
783	$combinedStyles = \array_merge($oldStyles, $newStyles);
784
785	$style = '';
786	foreach ($combinedStyles as $attributeName => $attributeValue) {
787	$style .= \strtolower(\trim($attributeName)) . ': ' . \trim($attributeValue) . '; ';
788	}
789	$trimmedStyle = \rtrim($style);
790
791	$this->caches[self::CACHE_KEY_COMBINED_STYLES][$cacheKey] = $trimmedStyle;
792
793	return $trimmedStyle;
794	}
795
796	/**
797	* Checks whether $attributeValue is marked as !important.
798	*
799	* @param string $attributeValue
800	*
801	* @return bool
802	*/
803	private function attributeValueIsImportant(string $attributeValue): bool
804	{
805	return (bool)\preg_match('/!\\s*+important$/i', $attributeValue);
806	}
807
808	/**
809	* Merges styles from styles attributes and style nodes and applies them to the attribute nodes
810	*/
811	private function fillStyleAttributesWithMergedStyles(): void
812	{
813	foreach ($this->styleAttributesForNodes as $nodePath => $styleAttributesForNode) {
814	$node = $this->visitedNodes[$nodePath];
815	$currentStyleAttributes = $this->parseCssDeclarationsBlock($node->getAttribute('style'));
816	$node->setAttribute(
817	'style',
818	$this->generateStyleStringFromDeclarationsArrays(
819	$currentStyleAttributes,
820	$styleAttributesForNode
821	)
822	);
823	}
824	}
825
826	/**
827	* Searches for all nodes with a style attribute and removes the "!important" annotations out of
828	* the inline style declarations, eventually by rearranging declarations.
829	*
830	* @throws \RuntimeException
831	*/
832	private function removeImportantAnnotationFromAllInlineStyles(): void
833	{
834	/** @var \DOMElement $node */
835	foreach ($this->getAllNodesWithStyleAttribute() as $node) {
836	$this->removeImportantAnnotationFromNodeInlineStyle($node);
837	}
838	}
839
840	/**
841	* Removes the "!important" annotations out of the inline style declarations,
842	* eventually by rearranging declarations.
843	* Rearranging needed when !important shorthand properties are followed by some of their
844	* not !important expanded-version properties.
845	* For example "font: 12px serif !important; font-size: 13px;" must be reordered
846	* to "font-size: 13px; font: 12px serif;" in order to remain correct.
847	*
848	* @param \DOMElement $node
849	*
850	* @throws \RuntimeException
851	*/
852	private function removeImportantAnnotationFromNodeInlineStyle(\DOMElement $node): void
853	{
854	$inlineStyleDeclarations = $this->parseCssDeclarationsBlock($node->getAttribute('style'));
855	/** @var array<string, string> $regularStyleDeclarations */
856	$regularStyleDeclarations = [];
857	/** @var array<string, string> $importantStyleDeclarations */
858	$importantStyleDeclarations = [];
859	foreach ($inlineStyleDeclarations as $property => $value) {
860	if ($this->attributeValueIsImportant($value)) {
861	$importantStyleDeclarations[$property] = $this->pregReplace('/\\s+!\\s+important$/i', '', $value);
862	} else {
863	$regularStyleDeclarations[$property] = $value;
864	}
865	}
866	$inlineStyleDeclarationsInNewOrder = \array_merge($regularStyleDeclarations, $importantStyleDeclarations);
867	$node->setAttribute(
868	'style',
869	$this->generateStyleStringFromSingleDeclarationsArray($inlineStyleDeclarationsInNewOrder)
870	);
871	}
872
873	/**
874	* Generates a CSS style string suitable to be used inline from the $styleDeclarations property => value array.
875	*
876	* @param array<string, string> $styleDeclarations
877	*
878	* @return string
879	*/
880	private function generateStyleStringFromSingleDeclarationsArray(array $styleDeclarations): string
881	{
882	return $this->generateStyleStringFromDeclarationsArrays([], $styleDeclarations);
883	}
884
885	/**
886	* Determines which of `$cssRules` actually apply to `$this->domDocument`, and sets them in
887	* `$this->matchingUninlinableCssRules`.
888	*
889	* @param array<array-key, array{
890	* media: string,
891	* selector: string,
892	* hasUnmatchablePseudo: bool,
893	* declarationsBlock: string,
894	* line: int
895	* }> $cssRules
896	* the "uninlinable" array of CSS rules returned by `collateCssRules`
897	*/
898	private function determineMatchingUninlinableCssRules(array $cssRules): void
899	{
900	$this->matchingUninlinableCssRules = \array_filter(
901	$cssRules,
902	function (array $cssRule): bool {
903	return $this->existsMatchForSelectorInCssRule($cssRule);
904	}
905	);
906	}
907
908	/**
909	* Checks whether there is at least one matching element for the CSS selector contained in the `selector` element
910	* of the provided CSS rule.
911	*
912	* Any dynamic pseudo-classes will be assumed to apply. If the selector matches a pseudo-element,
913	* it will test for a match with its originating element.
914	*
915	* @param array{
916	* media: string,
917	* selector: string,
918	* hasUnmatchablePseudo: bool,
919	* declarationsBlock: string,
920	* line: int
921	* } $cssRule
922	*
923	* @return bool
924	*
925	* @throws ParseException
926	*/
927	private function existsMatchForSelectorInCssRule(array $cssRule): bool
928	{
929	$selector = $cssRule['selector'];
930	if ($cssRule['hasUnmatchablePseudo']) {
931	$selector = $this->removeUnmatchablePseudoComponents($selector);
932	}
933	return $this->existsMatchForCssSelector($selector);
934	}
935
936	/**
937	* Checks whether there is at least one matching element for $cssSelector.
938	* When not in debug mode, it returns true also for invalid selectors (because they may be valid,
939	* just not implemented/recognized yet by Emogrifier).
940	*
941	* @param string $cssSelector
942	*
943	* @return bool
944	*
945	* @throws ParseException
946	*/
947	private function existsMatchForCssSelector(string $cssSelector): bool
948	{
949	try {
950	$nodesMatchingSelector = $this->getXPath()->query($this->getCssSelectorConverter()->toXPath($cssSelector));
951	} catch (ParseException $e) {
952	if ($this->debug) {
953	throw $e;
954	}
955	return true;
956	}
957
958	return $nodesMatchingSelector !== false && $nodesMatchingSelector->length !== 0;
959	}
960
961	/**
962	* Removes pseudo-elements and dynamic pseudo-classes from a CSS selector, replacing them with "*" if necessary.
963	* If such a pseudo-component is within the argument of `:not`, the entire `:not` component is removed or replaced.
964	*
965	* @param string $selector
966	*
967	* @return string
968	* selector which will match the relevant DOM elements if the pseudo-classes are assumed to apply, or in the
969	* case of pseudo-elements will match their originating element
970	*/
971	private function removeUnmatchablePseudoComponents(string $selector): string
972	{
973	// The regex allows nested brackets via `(?2)`.
974	// A space is temporarily prepended because the callback can't determine if the match was at the very start.
975	$selectorWithoutNots = \ltrim(\preg_replace_callback(
976	'/([\\s>+~]?+):not(\$[^()]+(?:(?2)[^()]+)*+\$)/i',
977	/** @param array<array-key, string> $matches */
978	function (array $matches): string {
979	return $this->replaceUnmatchableNotComponent($matches);
980	},
981	' ' . $selector
982	));
983
984	$selectorWithoutUnmatchablePseudoComponents = $this->removeSelectorComponents(
985	':(?!' . self::PSEUDO_CLASS_MATCHER . '):?+[\\w\\-]++(?:\$[^\$]*+\\))?+',
986	$selectorWithoutNots
987	);
988
989	if (
990	!\preg_match(
991	'/:(?:' . self::OF_TYPE_PSEUDO_CLASS_MATCHER . ')/i',
992	$selectorWithoutUnmatchablePseudoComponents
993	)
994	) {
995	return $selectorWithoutUnmatchablePseudoComponents;
996	}
997	return \implode('', \array_map(
998	function (string $selectorPart): string {
999	return $this->removeUnsupportedOfTypePseudoClasses($selectorPart);
1000	},
1001	\preg_split(
1002	'/(' . self::COMBINATOR_MATCHER . ')/',
1003	$selectorWithoutUnmatchablePseudoComponents,
1004	-1,
1005	PREG_SPLIT_DELIM_CAPTURE \| PREG_SPLIT_NO_EMPTY
1006	)
1007	));
1008	}
1009
1010	/**
1011	* Helps `removeUnmatchablePseudoComponents()` replace or remove a selector `:not(...)` component if its argument
1012	* contains pseudo-elements or dynamic pseudo-classes.
1013	*
1014	* @param array<array-key, string> $matches array of elements matched by the regular expression
1015	*
1016	* @return string
1017	* the full match if there were no unmatchable pseudo components within; otherwise, any preceding combinator
1018	* followed by "*", or an empty string if there was no preceding combinator
1019	*/
1020	private function replaceUnmatchableNotComponent(array $matches): string
1021	{
1022	[$notComponentWithAnyPrecedingCombinator, $anyPrecedingCombinator, $notArgumentInBrackets] = $matches;
1023
1024	if ($this->hasUnsupportedPseudoClass($notArgumentInBrackets)) {
1025	return $anyPrecedingCombinator !== '' ? $anyPrecedingCombinator . '*' : '';
1026	}
1027	return $notComponentWithAnyPrecedingCombinator;
1028	}
1029
1030	/**
1031	* Removes components from a CSS selector, replacing them with "*" if necessary.
1032	*
1033	* @param string $matcher regular expression part to match the components to remove
1034	* @param string $selector
1035	*
1036	* @return string
1037	* selector which will match the relevant DOM elements if the removed components are assumed to apply (or in
1038	* the case of pseudo-elements will match their originating element)
1039	*/
1040	private function removeSelectorComponents(string $matcher, string $selector): string
1041	{
1042	return \preg_replace(
1043	['/([\\s>+~]\|^)' . $matcher . '/i', '/' . $matcher . '/i'],
1044	['$1*', ''],
1045	$selector
1046	);
1047	}
1048
1049	/**
1050	* Removes any `...-of-type` pseudo-classes from part of a CSS selector, if it does not have a type, replacing them
1051	* with "*" if necessary.
1052	*
1053	* @param string $selectorPart part of a selector which has been split up at combinators
1054	*
1055	* @return string
1056	* selector part which will match the relevant DOM elements if the pseudo-classes are assumed to apply
1057	*/
1058	private function removeUnsupportedOfTypePseudoClasses(string $selectorPart): string
1059	{
1060	if (!$this->selectorPartHasUnsupportedOfTypePseudoClass($selectorPart)) {
1061	return $selectorPart;
1062	}
1063
1064	return $this->removeSelectorComponents(
1065	':(?:' . self::OF_TYPE_PSEUDO_CLASS_MATCHER . ')(?:\$[^\$]*+\\))?+',
1066	$selectorPart
1067	);
1068	}
1069
1070	/**
1071	* Applies `$this->matchingUninlinableCssRules` to `$this->domDocument` by placing them as CSS in a `<style>`
1072	* element.
1073	* If there are no uninlinable CSS rules to copy there, a `<style>` element will be created containing only the
1074	* applicable at-rules from `$parsedCss`.
1075	* If there are none of either, an empty `<style>` element will not be created.
1076	*
1077	* @param CssDocument $parsedCss
1078	* This may contain various at-rules whose content `CssInliner` does not currently attempt to inline or
1079	* process in any other way, such as `@import`, `@font-face`, `@keyframes`, etc., and which should precede
1080	* the processed but found-to-be-uninlinable CSS placed in the `<style>` element.
1081	* Note that `CssInliner` processes `@media` rules so that they can be ordered correctly with respect to
1082	* other uninlinable rules; these will not be duplicated from `$parsedCss`.
1083	*/
1084	private function copyUninlinableCssToStyleNode(CssDocument $parsedCss): void
1085	{
1086	$css = $parsedCss->renderNonConditionalAtRules();
1087
1088	// avoid including unneeded class dependency if there are no rules
1089	if ($this->getMatchingUninlinableCssRules() !== []) {
1090	$cssConcatenator = new CssConcatenator();
1091	foreach ($this->getMatchingUninlinableCssRules() as $cssRule) {
1092	$cssConcatenator->append([$cssRule['selector']], $cssRule['declarationsBlock'], $cssRule['media']);
1093	}
1094	$css .= $cssConcatenator->getCss();
1095	}
1096
1097	// avoid adding empty style element
1098	if ($css !== '') {
1099	$this->addStyleElementToDocument($css);
1100	}
1101	}
1102
1103	/**
1104	* Adds a style element with $css to $this->domDocument.
1105	*
1106	* This method is protected to allow overriding.
1107	*
1108	* @see https://github.com/MyIntervals/emogrifier/issues/103
1109	*
1110	* @param string $css
1111	*/
1112	protected function addStyleElementToDocument(string $css): void
1113	{
1114	$domDocument = $this->getDomDocument();
1115	$styleElement = $domDocument->createElement('style', $css);
1116	$styleAttribute = $domDocument->createAttribute('type');
1117	$styleAttribute->value = 'text/css';
1118	$styleElement->appendChild($styleAttribute);
1119
1120	$headElement = $this->getHeadElement();
1121	$headElement->appendChild($styleElement);
1122	}
1123
1124	/**
1125	* Returns the HEAD element.
1126	*
1127	* This method assumes that there always is a HEAD element.
1128	*
1129	* @return \DOMElement
1130	*
1131	* @throws \UnexpectedValueException
1132	*/
1133	private function getHeadElement(): \DOMElement
1134	{
1135	$node = $this->getDomDocument()->getElementsByTagName('head')->item(0);
1136	if (!$node instanceof \DOMElement) {
1137	throw new \UnexpectedValueException('There is no HEAD element. This should never happen.', 1617923227);
1138	}
1139
1140	return $node;
1141	}
1142
1143	/**
1144	* Wraps `preg_replace`. If an error occurs (which is highly unlikely), either it is logged and the original
1145	* `$subject` is returned, or in debug mode an exception is thrown.
1146	*
1147	* This method only supports strings, not arrays of strings.
1148	*
1149	* @param string $pattern
1150	* @param string $replacement
1151	* @param string $subject
1152	*
1153	* @return string
1154	*
1155	* @throws \RuntimeException
1156	*/
1157	private function pregReplace(string $pattern, string $replacement, string $subject): string
1158	{
1159	$result = \preg_replace($pattern, $replacement, $subject);
1160
1161	if (!\is_string($result)) {
1162	$this->logOrThrowPregLastError();
1163	$result = $subject;
1164	}
1165
1166	return $result;
1167	}
1168
1169	/**
1170	* Obtains the name of the error constant for `preg_last_error` (based on code posted at
1171	* {@see https://www.php.net/manual/en/function.preg-last-error.php#124124}) and puts it into an error message
1172	* which is either passed to `trigger_error` (in non-debug mode) or an exception which is thrown (in debug mode).
1173	*
1174	* @throws \RuntimeException
1175	*/
1176	private function logOrThrowPregLastError(): void
1177	{
1178	$pcreConstants = \get_defined_constants(true)['pcre'];
1179	$pcreErrorConstantNames = \array_flip(\array_filter(
1180	$pcreConstants,
1181	static function (string $key): bool {
1182	return \substr($key, -6) === '_ERROR';
1183	},
1184	ARRAY_FILTER_USE_KEY
1185	));
1186
1187	$pregLastError = \preg_last_error();
1188	$message = 'PCRE regex execution error `' . (string)($pcreErrorConstantNames[$pregLastError] ?? $pregLastError)
1189	. '`';
1190
1191	if ($this->debug) {
1192	throw new \RuntimeException($message, 1592870147);
1193	}
1194	\trigger_error($message);
1195	}
1196	}
1197