wp-user-avatar
/
third-party
/
vendor
/
pelago
/
emogrifier
/
src
/
HtmlProcessor
/
AbstractHtmlProcessor.php
AbstractHtmlProcessor.php
6 days ago
CssToAttributeConverter.php
6 days ago
CssVariableEvaluator.php
6 days ago
HtmlNormalizer.php
6 days ago
HtmlPruner.php
6 days ago
AbstractHtmlProcessor.php
383 lines
| 1 | <?php |
| 2 | |
| 3 | declare (strict_types=1); |
| 4 | namespace ProfilePressVendor\Pelago\Emogrifier\HtmlProcessor; |
| 5 | |
| 6 | use function ProfilePressVendor\Safe\preg_match; |
| 7 | use function ProfilePressVendor\Safe\preg_replace; |
| 8 | /** |
| 9 | * Base class for HTML processor that e.g., can remove, add or modify nodes or attributes. |
| 10 | * |
| 11 | * The "vanilla" subclass is the HtmlNormalizer. |
| 12 | */ |
| 13 | abstract class AbstractHtmlProcessor |
| 14 | { |
| 15 | protected const DEFAULT_DOCUMENT_TYPE = '<!DOCTYPE html>'; |
| 16 | protected const CONTENT_TYPE_META_TAG = '<meta http-equiv="Content-Type" content="text/html; charset=utf-8">'; |
| 17 | /** |
| 18 | * Regular expression part to match tag names that PHP's DOMDocument implementation is not |
| 19 | * aware are self-closing. These are mostly HTML5 elements, but for completeness `<command>` (obsolete) and |
| 20 | * `<keygen>` (deprecated) are also included. |
| 21 | * |
| 22 | * @see https://bugs.php.net/bug.php?id=73175 |
| 23 | */ |
| 24 | protected const PHP_UNRECOGNIZED_VOID_TAGNAME_MATCHER = '(?:command|embed|keygen|source|track|wbr)'; |
| 25 | /** |
| 26 | * Regular expression part to match tag names that may appear before the start of the `<body>` element. A start tag |
| 27 | * for any other element would implicitly start the `<body>` element due to tag omission rules. |
| 28 | */ |
| 29 | protected const TAGNAME_ALLOWED_BEFORE_BODY_MATCHER = '(?:html|head|base|command|link|meta|noscript|script|style|template|title)'; |
| 30 | /** |
| 31 | * regular expression pattern to match an HTML comment, including delimiters and modifiers |
| 32 | */ |
| 33 | protected const HTML_COMMENT_PATTERN = '/<!--[^-]*+(?:-(?!->)[^-]*+)*+(?:-->|$)/'; |
| 34 | /** |
| 35 | * regular expression pattern to match an HTML `<template>` element, including delimiters and modifiers |
| 36 | */ |
| 37 | protected const HTML_TEMPLATE_ELEMENT_PATTERN = '%<template[\s>][^<]*+(?:<(?!/template>)[^<]*+)*+(?:</template>|$)%i'; |
| 38 | /** |
| 39 | * @var \DOMDocument|null |
| 40 | */ |
| 41 | protected $domDocument = null; |
| 42 | /** |
| 43 | * @var \DOMXPath|null |
| 44 | */ |
| 45 | private $xPath = null; |
| 46 | /** |
| 47 | * The constructor. |
| 48 | * |
| 49 | * Please use `::fromHtml` or `::fromDomDocument` instead. |
| 50 | */ |
| 51 | private function __construct() |
| 52 | { |
| 53 | } |
| 54 | /** |
| 55 | * Builds a new instance from the given HTML. |
| 56 | * |
| 57 | * @param non-empty-string $unprocessedHtml raw HTML, must be UTF-encoded |
| 58 | * |
| 59 | * @return static |
| 60 | * |
| 61 | * @throws \InvalidArgumentException if $unprocessedHtml is anything other than a non-empty string |
| 62 | */ |
| 63 | public static function fromHtml(string $unprocessedHtml): self |
| 64 | { |
| 65 | // @phpstan-ignore-next-line argument.type We're checking for a contract violation here. |
| 66 | if ($unprocessedHtml === '') { |
| 67 | throw new \InvalidArgumentException('The provided HTML must not be empty.', 1515763647); |
| 68 | } |
| 69 | $instance = new static(); |
| 70 | $instance->setHtml($unprocessedHtml); |
| 71 | return $instance; |
| 72 | } |
| 73 | /** |
| 74 | * Builds a new instance from the given DOM document. |
| 75 | * |
| 76 | * @param \DOMDocument $document a DOM document returned by getDomDocument() of another instance |
| 77 | * |
| 78 | * @return static |
| 79 | */ |
| 80 | public static function fromDomDocument(\DOMDocument $document): self |
| 81 | { |
| 82 | $instance = new static(); |
| 83 | $instance->setDomDocument($document); |
| 84 | return $instance; |
| 85 | } |
| 86 | /** |
| 87 | * Sets the HTML to process. |
| 88 | * |
| 89 | * @param string $html the HTML to process, must be UTF-8-encoded |
| 90 | */ |
| 91 | private function setHtml(string $html): void |
| 92 | { |
| 93 | $this->createUnifiedDomDocument($html); |
| 94 | } |
| 95 | /** |
| 96 | * Provides access to the internal DOMDocument representation of the HTML in its current state. |
| 97 | * |
| 98 | * @throws \UnexpectedValueException |
| 99 | */ |
| 100 | public function getDomDocument(): \DOMDocument |
| 101 | { |
| 102 | if (!$this->domDocument instanceof \DOMDocument) { |
| 103 | $message = self::class . '::setDomDocument() has not yet been called on ' . static::class; |
| 104 | throw new \UnexpectedValueException($message, 1570472239); |
| 105 | } |
| 106 | return $this->domDocument; |
| 107 | } |
| 108 | private function setDomDocument(\DOMDocument $domDocument): void |
| 109 | { |
| 110 | $this->domDocument = $domDocument; |
| 111 | $this->xPath = new \DOMXPath($this->domDocument); |
| 112 | } |
| 113 | /** |
| 114 | * @throws \UnexpectedValueException |
| 115 | */ |
| 116 | protected function getXPath(): \DOMXPath |
| 117 | { |
| 118 | if (!$this->xPath instanceof \DOMXPath) { |
| 119 | $message = self::class . '::setDomDocument() has not yet been called on ' . static::class; |
| 120 | throw new \UnexpectedValueException($message, 1617819086); |
| 121 | } |
| 122 | return $this->xPath; |
| 123 | } |
| 124 | /** |
| 125 | * Renders the normalized and processed HTML. |
| 126 | */ |
| 127 | public function render(): string |
| 128 | { |
| 129 | $htmlWithPossibleErroneousClosingTags = $this->getDomDocument()->saveHTML(); |
| 130 | return $this->removeSelfClosingTagsClosingTags($htmlWithPossibleErroneousClosingTags); |
| 131 | } |
| 132 | /** |
| 133 | * Renders the content of the BODY element of the normalized and processed HTML. |
| 134 | */ |
| 135 | public function renderBodyContent(): string |
| 136 | { |
| 137 | $htmlWithPossibleErroneousClosingTags = $this->getDomDocument()->saveHTML($this->getBodyElement()); |
| 138 | $bodyNodeHtml = $this->removeSelfClosingTagsClosingTags($htmlWithPossibleErroneousClosingTags); |
| 139 | return preg_replace('%</?+body(?:\s[^>]*+)?+>%', '', $bodyNodeHtml); |
| 140 | } |
| 141 | /** |
| 142 | * Eliminates any invalid closing tags for void elements from the given HTML. |
| 143 | */ |
| 144 | private function removeSelfClosingTagsClosingTags(string $html): string |
| 145 | { |
| 146 | return preg_replace('%</' . self::PHP_UNRECOGNIZED_VOID_TAGNAME_MATCHER . '>%', '', $html); |
| 147 | } |
| 148 | /** |
| 149 | * Returns the HTML element. |
| 150 | * |
| 151 | * This method assumes that there always is an HTML element, throwing an exception otherwise. |
| 152 | * |
| 153 | * @throws \UnexpectedValueException |
| 154 | */ |
| 155 | protected function getHtmlElement(): \DOMElement |
| 156 | { |
| 157 | $htmlElement = $this->getDomDocument()->getElementsByTagName('html')->item(0); |
| 158 | if (!$htmlElement instanceof \DOMElement) { |
| 159 | throw new \UnexpectedValueException('There is no HTML element although there should be one.', 1569930853); |
| 160 | } |
| 161 | return $htmlElement; |
| 162 | } |
| 163 | /** |
| 164 | * Returns the BODY element. |
| 165 | * |
| 166 | * This method assumes that there always is a BODY element. |
| 167 | * |
| 168 | * @throws \RuntimeException |
| 169 | */ |
| 170 | private function getBodyElement(): \DOMElement |
| 171 | { |
| 172 | $node = $this->getDomDocument()->getElementsByTagName('body')->item(0); |
| 173 | if (!$node instanceof \DOMElement) { |
| 174 | throw new \RuntimeException('There is no body element.', 1617922607); |
| 175 | } |
| 176 | return $node; |
| 177 | } |
| 178 | /** |
| 179 | * Creates a DOM document from the given HTML and stores it in $this->domDocument. |
| 180 | * |
| 181 | * The DOM document will always have a BODY element and a document type. |
| 182 | */ |
| 183 | private function createUnifiedDomDocument(string $html): void |
| 184 | { |
| 185 | $this->createRawDomDocument($html); |
| 186 | $this->ensureExistenceOfBodyElement(); |
| 187 | } |
| 188 | /** |
| 189 | * Creates a DOMDocument instance from the given HTML and stores it in $this->domDocument. |
| 190 | */ |
| 191 | private function createRawDomDocument(string $html): void |
| 192 | { |
| 193 | $domDocument = new \DOMDocument(); |
| 194 | $domDocument->strictErrorChecking = \false; |
| 195 | $domDocument->formatOutput = \false; |
| 196 | $libXmlState = \libxml_use_internal_errors(\true); |
| 197 | $domDocument->loadHTML($this->prepareHtmlForDomConversion($html)); |
| 198 | \libxml_clear_errors(); |
| 199 | \libxml_use_internal_errors($libXmlState); |
| 200 | $this->setDomDocument($domDocument); |
| 201 | } |
| 202 | /** |
| 203 | * Returns the HTML with added document type, Content-Type meta tag, and self-closing slashes, if needed, |
| 204 | * ensuring that the HTML will be good for creating a DOM document from it. |
| 205 | */ |
| 206 | private function prepareHtmlForDomConversion(string $html): string |
| 207 | { |
| 208 | $htmlWithSelfClosingSlashes = $this->ensurePhpUnrecognizedSelfClosingTagsAreXml($html); |
| 209 | $htmlWithDocumentType = $this->ensureDocumentType($htmlWithSelfClosingSlashes); |
| 210 | return $this->addContentTypeMetaTag($htmlWithDocumentType); |
| 211 | } |
| 212 | /** |
| 213 | * Makes sure that the passed HTML has a document type, with lowercase "html". |
| 214 | * |
| 215 | * @return non-empty-string HTML with document type |
| 216 | */ |
| 217 | private function ensureDocumentType(string $html): string |
| 218 | { |
| 219 | $hasDocumentType = \stripos($html, '<!DOCTYPE') !== \false; |
| 220 | if ($hasDocumentType) { |
| 221 | return $this->normalizeDocumentType($html); |
| 222 | } |
| 223 | return self::DEFAULT_DOCUMENT_TYPE . $html; |
| 224 | } |
| 225 | /** |
| 226 | * Makes sure the document type in the passed HTML has lowercase `html`. |
| 227 | * |
| 228 | * @param non-empty-string $html |
| 229 | * |
| 230 | * @return non-empty-string HTML with normalized document type |
| 231 | */ |
| 232 | private function normalizeDocumentType(string $html): string |
| 233 | { |
| 234 | // Limit to replacing the first occurrence: as an optimization; and in case an example exists as unescaped text. |
| 235 | $result = preg_replace('/<!DOCTYPE\s++html(?=[\s>])/i', '<!DOCTYPE html', $html, 1); |
| 236 | \assert($result !== ''); |
| 237 | return $result; |
| 238 | } |
| 239 | /** |
| 240 | * Adds a Content-Type meta tag for the charset. |
| 241 | * |
| 242 | * This method also ensures that there is a HEAD element. |
| 243 | * |
| 244 | * @param non-empty-string $html |
| 245 | * |
| 246 | * @return non-empty-string |
| 247 | */ |
| 248 | private function addContentTypeMetaTag(string $html): string |
| 249 | { |
| 250 | if ($this->hasContentTypeMetaTagInHead($html)) { |
| 251 | return $html; |
| 252 | } |
| 253 | // We are trying to insert the meta tag to the right spot in the DOM. |
| 254 | // If we just prepended it to the HTML, we would lose attributes set to the HTML tag. |
| 255 | $hasHeadTag = preg_match('/<head[\s>]/i', $html) !== 0; |
| 256 | $hasHtmlTag = \stripos($html, '<html') !== \false; |
| 257 | if ($hasHeadTag) { |
| 258 | $reworkedHtml = preg_replace('/<head(?=[\s>])([^>]*+)>/i', '<head$1>' . self::CONTENT_TYPE_META_TAG, $html); |
| 259 | } elseif ($hasHtmlTag) { |
| 260 | $reworkedHtml = preg_replace('/<html(.*?)>/is', '<html$1><head>' . self::CONTENT_TYPE_META_TAG . '</head>', $html); |
| 261 | } else { |
| 262 | $reworkedHtml = self::CONTENT_TYPE_META_TAG . $html; |
| 263 | } |
| 264 | \assert($reworkedHtml !== ''); |
| 265 | return $reworkedHtml; |
| 266 | } |
| 267 | /** |
| 268 | * Tests whether the given HTML has a valid `Content-Type` metadata element within the `<head>` element. Due to tag |
| 269 | * omission rules, HTML parsers are expected to end the `<head>` element and start the `<body>` element upon |
| 270 | * encountering a start tag for any element which is permitted only within the `<body>`. |
| 271 | */ |
| 272 | private function hasContentTypeMetaTagInHead(string $html): bool |
| 273 | { |
| 274 | preg_match('% |
| 275 | (?(DEFINE) |
| 276 | # the target `http-equiv` attribute match |
| 277 | (?<target_attribute> |
| 278 | http-equiv=(["\']?+)Content-Type\g{-1} |
| 279 | # must be followed by one of these characters |
| 280 | [\s/>] |
| 281 | ) |
| 282 | # the target `meta` element match without the opening `<` |
| 283 | (?<target> |
| 284 | meta(?=\s) |
| 285 | # one or other of these |
| 286 | (?: |
| 287 | # one or more characters other than `>` or space |
| 288 | [^>\s]++ |
| 289 | | |
| 290 | # space not followed by the target `http-equiv` attribute |
| 291 | \s(?!(?&target_attribute)) |
| 292 | ) |
| 293 | # any number of times (including zero) |
| 294 | *+ |
| 295 | \s(?&target_attribute) |
| 296 | ) |
| 297 | ) |
| 298 | # start of `subject` |
| 299 | ^ |
| 300 | # one or other of these |
| 301 | (?: |
| 302 | # one or more characters other than `<` |
| 303 | [^<]++ |
| 304 | | |
| 305 | # `<` not followed by `target` |
| 306 | <(?!(?&target)) |
| 307 | ) |
| 308 | # any number of times (including zero) |
| 309 | *+ |
| 310 | # followed by the target, not captured |
| 311 | (?=<(?&target)) |
| 312 | %isx', $html, $matches); |
| 313 | if (isset($matches[0])) { |
| 314 | $htmlBefore = $matches[0]; |
| 315 | try { |
| 316 | $hasContentTypeMetaTagInHead = !$this->hasEndOfHeadElement($htmlBefore); |
| 317 | } catch (\RuntimeException $exception) { |
| 318 | // If something unexpected occurs, assume the `Content-Type` that was found is valid. |
| 319 | \trigger_error($exception->getMessage()); |
| 320 | $hasContentTypeMetaTagInHead = \true; |
| 321 | } |
| 322 | } else { |
| 323 | $hasContentTypeMetaTagInHead = \false; |
| 324 | } |
| 325 | return $hasContentTypeMetaTagInHead; |
| 326 | } |
| 327 | /** |
| 328 | * Tests whether the `<head>` element ends within the given HTML. Due to tag omission rules, HTML parsers are |
| 329 | * expected to end the `<head>` element and start the `<body>` element upon encountering a start tag for any element |
| 330 | * which is permitted only within the `<body>`. |
| 331 | * |
| 332 | * @throws \RuntimeException |
| 333 | */ |
| 334 | private function hasEndOfHeadElement(string $html): bool |
| 335 | { |
| 336 | if (preg_match('%<(?!' . self::TAGNAME_ALLOWED_BEFORE_BODY_MATCHER . '[\s/>])\w|</head>%i', $html) !== 0) { |
| 337 | // An exception to the implicit end of the `<head>` is any content within a `<template>` element, as well in |
| 338 | // comments. As an optimization, this is only checked for if a potential `<head>` end tag is found. |
| 339 | $htmlWithoutCommentsOrTemplates = $this->removeHtmlTemplateElements($this->removeHtmlComments($html)); |
| 340 | $hasEndOfHeadElement = $htmlWithoutCommentsOrTemplates === $html || $this->hasEndOfHeadElement($htmlWithoutCommentsOrTemplates); |
| 341 | } else { |
| 342 | $hasEndOfHeadElement = \false; |
| 343 | } |
| 344 | return $hasEndOfHeadElement; |
| 345 | } |
| 346 | /** |
| 347 | * Removes comments from the given HTML, including any which are unterminated, for which the remainder of the string |
| 348 | * is removed. |
| 349 | */ |
| 350 | private function removeHtmlComments(string $html): string |
| 351 | { |
| 352 | return preg_replace(self::HTML_COMMENT_PATTERN, '', $html); |
| 353 | } |
| 354 | /** |
| 355 | * Removes `<template>` elements from the given HTML, including any without an end tag, for which the remainder of |
| 356 | * the string is removed. |
| 357 | */ |
| 358 | private function removeHtmlTemplateElements(string $html): string |
| 359 | { |
| 360 | return preg_replace(self::HTML_TEMPLATE_ELEMENT_PATTERN, '', $html); |
| 361 | } |
| 362 | /** |
| 363 | * Makes sure that any self-closing tags not recognized as such by PHP's DOMDocument implementation have a |
| 364 | * self-closing slash. |
| 365 | */ |
| 366 | private function ensurePhpUnrecognizedSelfClosingTagsAreXml(string $html): string |
| 367 | { |
| 368 | return preg_replace('%<' . self::PHP_UNRECOGNIZED_VOID_TAGNAME_MATCHER . '\b[^>]*+(?<!/)(?=>)%', '$0/', $html); |
| 369 | } |
| 370 | /** |
| 371 | * Checks that $this->domDocument has a BODY element and adds it if it is missing. |
| 372 | * |
| 373 | * @throws \UnexpectedValueException |
| 374 | */ |
| 375 | private function ensureExistenceOfBodyElement(): void |
| 376 | { |
| 377 | if ($this->getDomDocument()->getElementsByTagName('body')->item(0) instanceof \DOMElement) { |
| 378 | return; |
| 379 | } |
| 380 | $this->getHtmlElement()->appendChild($this->getDomDocument()->createElement('body')); |
| 381 | } |
| 382 | } |
| 383 |