HTML5Entities.php
3 years ago
OutputRules.php
3 years ago
RulesInterface.php
3 years ago
Traverser.php
3 years ago
OutputRules.php
390 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * @file |
| 5 | * The rules for generating output in the serializer. |
| 6 | * |
| 7 | * These output rules are likely to generate output similar to the document that |
| 8 | * was parsed. It is not intended to output exactly the document that was parsed. |
| 9 | */ |
| 10 | namespace IAWP\Masterminds\HTML5\Serializer; |
| 11 | |
| 12 | use IAWP\Masterminds\HTML5\Elements; |
| 13 | /** |
| 14 | * Generate the output html5 based on element rules. |
| 15 | */ |
| 16 | class OutputRules implements RulesInterface |
| 17 | { |
| 18 | /** |
| 19 | * Defined in http://www.w3.org/TR/html51/infrastructure.html#html-namespace-0. |
| 20 | */ |
| 21 | const NAMESPACE_HTML = 'http://www.w3.org/1999/xhtml'; |
| 22 | const NAMESPACE_MATHML = 'http://www.w3.org/1998/Math/MathML'; |
| 23 | const NAMESPACE_SVG = 'http://www.w3.org/2000/svg'; |
| 24 | const NAMESPACE_XLINK = 'http://www.w3.org/1999/xlink'; |
| 25 | const NAMESPACE_XML = 'http://www.w3.org/XML/1998/namespace'; |
| 26 | const NAMESPACE_XMLNS = 'http://www.w3.org/2000/xmlns/'; |
| 27 | /** |
| 28 | * Holds the HTML5 element names that causes a namespace switch. |
| 29 | * |
| 30 | * @var array |
| 31 | */ |
| 32 | protected $implicitNamespaces = array(self::NAMESPACE_HTML, self::NAMESPACE_SVG, self::NAMESPACE_MATHML, self::NAMESPACE_XML, self::NAMESPACE_XMLNS); |
| 33 | const IM_IN_HTML = 1; |
| 34 | const IM_IN_SVG = 2; |
| 35 | const IM_IN_MATHML = 3; |
| 36 | /** |
| 37 | * Used as cache to detect if is available ENT_HTML5. |
| 38 | * |
| 39 | * @var bool |
| 40 | */ |
| 41 | private $hasHTML5 = \false; |
| 42 | protected $traverser; |
| 43 | protected $encode = \false; |
| 44 | protected $out; |
| 45 | protected $outputMode; |
| 46 | private $xpath; |
| 47 | protected $nonBooleanAttributes = array( |
| 48 | /* |
| 49 | array( |
| 50 | 'nodeNamespace'=>'http://www.w3.org/1999/xhtml', |
| 51 | 'attrNamespace'=>'http://www.w3.org/1999/xhtml', |
| 52 | |
| 53 | 'nodeName'=>'img', 'nodeName'=>array('img', 'a'), |
| 54 | 'attrName'=>'alt', 'attrName'=>array('title', 'alt'), |
| 55 | ), |
| 56 | */ |
| 57 | array('nodeNamespace' => 'http://www.w3.org/1999/xhtml', 'attrName' => array('href', 'hreflang', 'http-equiv', 'icon', 'id', 'keytype', 'kind', 'label', 'lang', 'language', 'list', 'maxlength', 'media', 'method', 'name', 'placeholder', 'rel', 'rows', 'rowspan', 'sandbox', 'spellcheck', 'scope', 'seamless', 'shape', 'size', 'sizes', 'span', 'src', 'srcdoc', 'srclang', 'srcset', 'start', 'step', 'style', 'summary', 'tabindex', 'target', 'title', 'type', 'value', 'width', 'border', 'charset', 'cite', 'class', 'code', 'codebase', 'color', 'cols', 'colspan', 'content', 'coords', 'data', 'datetime', 'default', 'dir', 'dirname', 'enctype', 'for', 'form', 'formaction', 'headers', 'height', 'accept', 'accept-charset', 'accesskey', 'action', 'align', 'alt', 'bgcolor')), |
| 58 | array('nodeNamespace' => 'http://www.w3.org/1999/xhtml', 'xpath' => 'starts-with(local-name(), \'data-\')'), |
| 59 | ); |
| 60 | const DOCTYPE = '<!DOCTYPE html>'; |
| 61 | public function __construct($output, $options = array()) |
| 62 | { |
| 63 | if (isset($options['encode_entities'])) { |
| 64 | $this->encode = $options['encode_entities']; |
| 65 | } |
| 66 | $this->outputMode = static::IM_IN_HTML; |
| 67 | $this->out = $output; |
| 68 | $this->hasHTML5 = \defined('ENT_HTML5'); |
| 69 | } |
| 70 | public function addRule(array $rule) |
| 71 | { |
| 72 | $this->nonBooleanAttributes[] = $rule; |
| 73 | } |
| 74 | public function setTraverser(Traverser $traverser) |
| 75 | { |
| 76 | $this->traverser = $traverser; |
| 77 | return $this; |
| 78 | } |
| 79 | public function unsetTraverser() |
| 80 | { |
| 81 | $this->traverser = null; |
| 82 | return $this; |
| 83 | } |
| 84 | public function document($dom) |
| 85 | { |
| 86 | $this->doctype(); |
| 87 | if ($dom->documentElement) { |
| 88 | foreach ($dom->childNodes as $node) { |
| 89 | $this->traverser->node($node); |
| 90 | } |
| 91 | $this->nl(); |
| 92 | } |
| 93 | } |
| 94 | protected function doctype() |
| 95 | { |
| 96 | $this->wr(static::DOCTYPE); |
| 97 | $this->nl(); |
| 98 | } |
| 99 | public function element($ele) |
| 100 | { |
| 101 | $name = $ele->tagName; |
| 102 | // Per spec: |
| 103 | // If the element has a declared namespace in the HTML, MathML or |
| 104 | // SVG namespaces, we use the lname instead of the tagName. |
| 105 | if ($this->traverser->isLocalElement($ele)) { |
| 106 | $name = $ele->localName; |
| 107 | } |
| 108 | // If we are in SVG or MathML there is special handling. |
| 109 | // Using if/elseif instead of switch because it's faster in PHP. |
| 110 | if ('svg' == $name) { |
| 111 | $this->outputMode = static::IM_IN_SVG; |
| 112 | $name = Elements::normalizeSvgElement($name); |
| 113 | } elseif ('math' == $name) { |
| 114 | $this->outputMode = static::IM_IN_MATHML; |
| 115 | } |
| 116 | $this->openTag($ele); |
| 117 | if (Elements::isA($name, Elements::TEXT_RAW)) { |
| 118 | foreach ($ele->childNodes as $child) { |
| 119 | if ($child instanceof \DOMCharacterData) { |
| 120 | $this->wr($child->data); |
| 121 | } elseif ($child instanceof \DOMElement) { |
| 122 | $this->element($child); |
| 123 | } |
| 124 | } |
| 125 | } else { |
| 126 | // Handle children. |
| 127 | if ($ele->hasChildNodes()) { |
| 128 | $this->traverser->children($ele->childNodes); |
| 129 | } |
| 130 | // Close out the SVG or MathML special handling. |
| 131 | if ('svg' == $name || 'math' == $name) { |
| 132 | $this->outputMode = static::IM_IN_HTML; |
| 133 | } |
| 134 | } |
| 135 | // If not unary, add a closing tag. |
| 136 | if (!Elements::isA($name, Elements::VOID_TAG)) { |
| 137 | $this->closeTag($ele); |
| 138 | } |
| 139 | } |
| 140 | /** |
| 141 | * Write a text node. |
| 142 | * |
| 143 | * @param \DOMText $ele The text node to write. |
| 144 | */ |
| 145 | public function text($ele) |
| 146 | { |
| 147 | if (isset($ele->parentNode) && isset($ele->parentNode->tagName) && Elements::isA($ele->parentNode->localName, Elements::TEXT_RAW)) { |
| 148 | $this->wr($ele->data); |
| 149 | return; |
| 150 | } |
| 151 | // FIXME: This probably needs some flags set. |
| 152 | $this->wr($this->enc($ele->data)); |
| 153 | } |
| 154 | public function cdata($ele) |
| 155 | { |
| 156 | // This encodes CDATA. |
| 157 | $this->wr($ele->ownerDocument->saveXML($ele)); |
| 158 | } |
| 159 | public function comment($ele) |
| 160 | { |
| 161 | // These produce identical output. |
| 162 | // $this->wr('<!--')->wr($ele->data)->wr('-->'); |
| 163 | $this->wr($ele->ownerDocument->saveXML($ele)); |
| 164 | } |
| 165 | public function processorInstruction($ele) |
| 166 | { |
| 167 | $this->wr('<?')->wr($ele->target)->wr(' ')->wr($ele->data)->wr('?>'); |
| 168 | } |
| 169 | /** |
| 170 | * Write the namespace attributes. |
| 171 | * |
| 172 | * @param \DOMNode $ele The element being written. |
| 173 | */ |
| 174 | protected function namespaceAttrs($ele) |
| 175 | { |
| 176 | if (!$this->xpath || $this->xpath->document !== $ele->ownerDocument) { |
| 177 | $this->xpath = new \DOMXPath($ele->ownerDocument); |
| 178 | } |
| 179 | foreach ($this->xpath->query('namespace::*[not(.=../../namespace::*)]', $ele) as $nsNode) { |
| 180 | if (!\in_array($nsNode->nodeValue, $this->implicitNamespaces)) { |
| 181 | $this->wr(' ')->wr($nsNode->nodeName)->wr('="')->wr($nsNode->nodeValue)->wr('"'); |
| 182 | } |
| 183 | } |
| 184 | } |
| 185 | /** |
| 186 | * Write the opening tag. |
| 187 | * |
| 188 | * Tags for HTML, MathML, and SVG are in the local name. Otherwise, use the |
| 189 | * qualified name (8.3). |
| 190 | * |
| 191 | * @param \DOMNode $ele The element being written. |
| 192 | */ |
| 193 | protected function openTag($ele) |
| 194 | { |
| 195 | $this->wr('<')->wr($this->traverser->isLocalElement($ele) ? $ele->localName : $ele->tagName); |
| 196 | $this->attrs($ele); |
| 197 | $this->namespaceAttrs($ele); |
| 198 | if ($this->outputMode == static::IM_IN_HTML) { |
| 199 | $this->wr('>'); |
| 200 | } else { |
| 201 | if ($ele->hasChildNodes()) { |
| 202 | $this->wr('>'); |
| 203 | } else { |
| 204 | $this->wr(' />'); |
| 205 | } |
| 206 | } |
| 207 | } |
| 208 | protected function attrs($ele) |
| 209 | { |
| 210 | // FIXME: Needs support for xml, xmlns, xlink, and namespaced elements. |
| 211 | if (!$ele->hasAttributes()) { |
| 212 | return $this; |
| 213 | } |
| 214 | // TODO: Currently, this always writes name="value", and does not do |
| 215 | // value-less attributes. |
| 216 | $map = $ele->attributes; |
| 217 | $len = $map->length; |
| 218 | for ($i = 0; $i < $len; ++$i) { |
| 219 | $node = $map->item($i); |
| 220 | $val = $this->enc($node->value, \true); |
| 221 | // XXX: The spec says that we need to ensure that anything in |
| 222 | // the XML, XMLNS, or XLink NS's should use the canonical |
| 223 | // prefix. It seems that DOM does this for us already, but there |
| 224 | // may be exceptions. |
| 225 | $name = $node->nodeName; |
| 226 | // Special handling for attributes in SVG and MathML. |
| 227 | // Using if/elseif instead of switch because it's faster in PHP. |
| 228 | if ($this->outputMode == static::IM_IN_SVG) { |
| 229 | $name = Elements::normalizeSvgAttribute($name); |
| 230 | } elseif ($this->outputMode == static::IM_IN_MATHML) { |
| 231 | $name = Elements::normalizeMathMlAttribute($name); |
| 232 | } |
| 233 | $this->wr(' ')->wr($name); |
| 234 | if (isset($val) && '' !== $val || $this->nonBooleanAttribute($node)) { |
| 235 | $this->wr('="')->wr($val)->wr('"'); |
| 236 | } |
| 237 | } |
| 238 | } |
| 239 | protected function nonBooleanAttribute(\DOMAttr $attr) |
| 240 | { |
| 241 | $ele = $attr->ownerElement; |
| 242 | foreach ($this->nonBooleanAttributes as $rule) { |
| 243 | if (isset($rule['nodeNamespace']) && $rule['nodeNamespace'] !== $ele->namespaceURI) { |
| 244 | continue; |
| 245 | } |
| 246 | if (isset($rule['attNamespace']) && $rule['attNamespace'] !== $attr->namespaceURI) { |
| 247 | continue; |
| 248 | } |
| 249 | if (isset($rule['nodeName']) && !\is_array($rule['nodeName']) && $rule['nodeName'] !== $ele->localName) { |
| 250 | continue; |
| 251 | } |
| 252 | if (isset($rule['nodeName']) && \is_array($rule['nodeName']) && !\in_array($ele->localName, $rule['nodeName'], \true)) { |
| 253 | continue; |
| 254 | } |
| 255 | if (isset($rule['attrName']) && !\is_array($rule['attrName']) && $rule['attrName'] !== $attr->localName) { |
| 256 | continue; |
| 257 | } |
| 258 | if (isset($rule['attrName']) && \is_array($rule['attrName']) && !\in_array($attr->localName, $rule['attrName'], \true)) { |
| 259 | continue; |
| 260 | } |
| 261 | if (isset($rule['xpath'])) { |
| 262 | $xp = $this->getXPath($attr); |
| 263 | if (isset($rule['prefixes'])) { |
| 264 | foreach ($rule['prefixes'] as $nsPrefix => $ns) { |
| 265 | $xp->registerNamespace($nsPrefix, $ns); |
| 266 | } |
| 267 | } |
| 268 | if (!$xp->evaluate($rule['xpath'], $attr)) { |
| 269 | continue; |
| 270 | } |
| 271 | } |
| 272 | return \true; |
| 273 | } |
| 274 | return \false; |
| 275 | } |
| 276 | private function getXPath(\DOMNode $node) |
| 277 | { |
| 278 | if (!$this->xpath) { |
| 279 | $this->xpath = new \DOMXPath($node->ownerDocument); |
| 280 | } |
| 281 | return $this->xpath; |
| 282 | } |
| 283 | /** |
| 284 | * Write the closing tag. |
| 285 | * |
| 286 | * Tags for HTML, MathML, and SVG are in the local name. Otherwise, use the |
| 287 | * qualified name (8.3). |
| 288 | * |
| 289 | * @param \DOMNode $ele The element being written. |
| 290 | */ |
| 291 | protected function closeTag($ele) |
| 292 | { |
| 293 | if ($this->outputMode == static::IM_IN_HTML || $ele->hasChildNodes()) { |
| 294 | $this->wr('</')->wr($this->traverser->isLocalElement($ele) ? $ele->localName : $ele->tagName)->wr('>'); |
| 295 | } |
| 296 | } |
| 297 | /** |
| 298 | * Write to the output. |
| 299 | * |
| 300 | * @param string $text The string to put into the output |
| 301 | * |
| 302 | * @return $this |
| 303 | */ |
| 304 | protected function wr($text) |
| 305 | { |
| 306 | \fwrite($this->out, $text); |
| 307 | return $this; |
| 308 | } |
| 309 | /** |
| 310 | * Write a new line character. |
| 311 | * |
| 312 | * @return $this |
| 313 | */ |
| 314 | protected function nl() |
| 315 | { |
| 316 | \fwrite($this->out, \PHP_EOL); |
| 317 | return $this; |
| 318 | } |
| 319 | /** |
| 320 | * Encode text. |
| 321 | * |
| 322 | * When encode is set to false, the default value, the text passed in is |
| 323 | * escaped per section 8.3 of the html5 spec. For details on how text is |
| 324 | * escaped see the escape() method. |
| 325 | * |
| 326 | * When encoding is set to true the text is converted to named character |
| 327 | * references where appropriate. Section 8.1.4 Character references of the |
| 328 | * html5 spec refers to using named character references. This is useful for |
| 329 | * characters that can't otherwise legally be used in the text. |
| 330 | * |
| 331 | * The named character references are listed in section 8.5. |
| 332 | * |
| 333 | * @see http://www.w3.org/TR/2013/CR-html5-20130806/syntax.html#named-character-references True encoding will turn all named character references into their entities. |
| 334 | * This includes such characters as +.# and many other common ones. By default |
| 335 | * encoding here will just escape &'<>". |
| 336 | * |
| 337 | * Note, PHP 5.4+ has better html5 encoding. |
| 338 | * |
| 339 | * @todo Use the Entities class in php 5.3 to have html5 entities. |
| 340 | * |
| 341 | * @param string $text Text to encode. |
| 342 | * @param bool $attribute True if we are encoding an attrubute, false otherwise. |
| 343 | * |
| 344 | * @return string The encoded text. |
| 345 | */ |
| 346 | protected function enc($text, $attribute = \false) |
| 347 | { |
| 348 | // Escape the text rather than convert to named character references. |
| 349 | if (!$this->encode) { |
| 350 | return $this->escape($text, $attribute); |
| 351 | } |
| 352 | // If we are in PHP 5.4+ we can use the native html5 entity functionality to |
| 353 | // convert the named character references. |
| 354 | if ($this->hasHTML5) { |
| 355 | return \htmlentities($text, \ENT_HTML5 | \ENT_SUBSTITUTE | \ENT_QUOTES, 'UTF-8', \false); |
| 356 | } else { |
| 357 | return \strtr($text, HTML5Entities::$map); |
| 358 | } |
| 359 | } |
| 360 | /** |
| 361 | * Escape test. |
| 362 | * |
| 363 | * According to the html5 spec section 8.3 Serializing HTML fragments, text |
| 364 | * within tags that are not style, script, xmp, iframe, noembed, and noframes |
| 365 | * need to be properly escaped. |
| 366 | * |
| 367 | * The & should be converted to &, no breaking space unicode characters |
| 368 | * converted to , when in attribute mode the " should be converted to |
| 369 | * ", and when not in attribute mode the < and > should be converted to |
| 370 | * < and >. |
| 371 | * |
| 372 | * @see http://www.w3.org/TR/2013/CR-html5-20130806/syntax.html#escapingString |
| 373 | * |
| 374 | * @param string $text Text to escape. |
| 375 | * @param bool $attribute True if we are escaping an attrubute, false otherwise. |
| 376 | */ |
| 377 | protected function escape($text, $attribute = \false) |
| 378 | { |
| 379 | // Not using htmlspecialchars because, while it does escaping, it doesn't |
| 380 | // match the requirements of section 8.5. For example, it doesn't handle |
| 381 | // non-breaking spaces. |
| 382 | if ($attribute) { |
| 383 | $replace = array('"' => '"', '&' => '&', " " => ' '); |
| 384 | } else { |
| 385 | $replace = array('<' => '<', '>' => '>', '&' => '&', " " => ' '); |
| 386 | } |
| 387 | return \strtr($text, $replace); |
| 388 | } |
| 389 | } |
| 390 |