wp-parsely / vendor / masterminds / html5 / src / HTML5 / Elements.php

Elements.php in Parse.ly 3.24.2, at vendor/masterminds/html5/src/HTML5/Elements.php

640 lines 20.1 KB

Raw Download Zip

1	<?php
2	/**
3	* Provide general element functions.
4	*/
5
6	namespace Masterminds\HTML5;
7
8	/**
9	* This class provides general information about HTML5 elements,
10	* including syntactic and semantic issues.
11	* Parsers and serializers can
12	* use this class as a reference point for information about the rules
13	* of various HTML5 elements.
14	*
15	* @todo consider using a bitmask table lookup. There is enough overlap in
16	* naming that this could significantly shrink the size and maybe make it
17	* faster. See the Go teams implementation at https://code.google.com/p/go/source/browse/html/atom.
18	*/
19	class Elements
20	{
21	/**
22	* Indicates an element is described in the specification.
23	*/
24	const KNOWN_ELEMENT = 1;
25
26	// From section 8.1.2: "script", "style"
27	// From 8.2.5.4.7 ("in body" insertion mode): "noembed"
28	// From 8.4 "style", "xmp", "iframe", "noembed", "noframes"
29	/**
30	* Indicates the contained text should be processed as raw text.
31	*/
32	const TEXT_RAW = 2;
33
34	// From section 8.1.2: "textarea", "title"
35	/**
36	* Indicates the contained text should be processed as RCDATA.
37	*/
38	const TEXT_RCDATA = 4;
39
40	/**
41	* Indicates the tag cannot have content.
42	*/
43	const VOID_TAG = 8;
44
45	// "address", "article", "aside", "blockquote", "center", "details", "dialog", "dir", "div", "dl",
46	// "fieldset", "figcaption", "figure", "footer", "header", "hgroup", "menu",
47	// "nav", "ol", "p", "section", "summary", "ul"
48	// "h1", "h2", "h3", "h4", "h5", "h6"
49	// "pre", "listing"
50	// "form"
51	// "plaintext"
52	/**
53	* Indicates that if a previous event is for a P tag, that element
54	* should be considered closed.
55	*/
56	const AUTOCLOSE_P = 16;
57
58	/**
59	* Indicates that the text inside is plaintext (pre).
60	*/
61	const TEXT_PLAINTEXT = 32;
62
63	// See https://developer.mozilla.org/en-US/docs/HTML/Block-level_elements
64	/**
65	* Indicates that the tag is a block.
66	*/
67	const BLOCK_TAG = 64;
68
69	/**
70	* Indicates that the tag allows only inline elements as child nodes.
71	*/
72	const BLOCK_ONLY_INLINE = 128;
73
74	/**
75	* Elements with optional end tags that cause auto-closing of previous and parent tags,
76	* as example most of the table related tags, see https://www.w3.org/TR/html401/struct/tables.html
77	* Structure is as follows:
78	* TAG-NAME => [PARENT-TAG-NAME-TO-CLOSE1, PARENT-TAG-NAME-TO-CLOSE2, ...].
79	*
80	* Order is important, after auto-closing one parent with might have to close also their parent.
81	*
82	* @var array<string, string[]>
83	*/
84	public static $optionalEndElementsParentsToClose = array(
85	'tr' => array('td', 'tr'),
86	'td' => array('td', 'th'),
87	'th' => array('td', 'th'),
88	'tfoot' => array('td', 'th', 'tr', 'tbody', 'thead'),
89	'tbody' => array('td', 'th', 'tr', 'thead'),
90	);
91
92	/**
93	* The HTML5 elements as defined in http://dev.w3.org/html5/markup/elements.html.
94	*
95	* @var array
96	*/
97	public static $html5 = array(
98	'a' => 1,
99	'abbr' => 1,
100	'address' => 65, // NORMAL \| BLOCK_TAG
101	'area' => 9, // NORMAL \| VOID_TAG
102	'article' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
103	'aside' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
104	'audio' => 1, // NORMAL
105	'b' => 1,
106	'base' => 9, // NORMAL \| VOID_TAG
107	'bdi' => 1,
108	'bdo' => 1,
109	'blockquote' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
110	'body' => 1,
111	'br' => 9, // NORMAL \| VOID_TAG
112	'button' => 1,
113	'canvas' => 65, // NORMAL \| BLOCK_TAG
114	'caption' => 1,
115	'cite' => 1,
116	'code' => 1,
117	'col' => 9, // NORMAL \| VOID_TAG
118	'colgroup' => 1,
119	'command' => 9, // NORMAL \| VOID_TAG
120	// "data" => 1, // This is highly experimental and only part of the whatwg spec (not w3c). See https://developer.mozilla.org/en-US/docs/HTML/Element/data
121	'datalist' => 1,
122	'dd' => 65, // NORMAL \| BLOCK_TAG
123	'del' => 1,
124	'details' => 17, // NORMAL \| AUTOCLOSE_P,
125	'dfn' => 1,
126	'dialog' => 17, // NORMAL \| AUTOCLOSE_P,
127	'div' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
128	'dl' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
129	'dt' => 1,
130	'em' => 1,
131	'embed' => 9, // NORMAL \| VOID_TAG
132	'fieldset' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
133	'figcaption' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
134	'figure' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
135	'footer' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
136	'form' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
137	'h1' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
138	'h2' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
139	'h3' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
140	'h4' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
141	'h5' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
142	'h6' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
143	'head' => 1,
144	'header' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
145	'hgroup' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
146	'hr' => 73, // NORMAL \| VOID_TAG
147	'html' => 1,
148	'i' => 1,
149	'iframe' => 3, // NORMAL \| TEXT_RAW
150	'img' => 9, // NORMAL \| VOID_TAG
151	'input' => 9, // NORMAL \| VOID_TAG
152	'kbd' => 1,
153	'ins' => 1,
154	'keygen' => 9, // NORMAL \| VOID_TAG
155	'label' => 1,
156	'legend' => 1,
157	'li' => 1,
158	'link' => 9, // NORMAL \| VOID_TAG
159	'map' => 1,
160	'mark' => 1,
161	'menu' => 17, // NORMAL \| AUTOCLOSE_P,
162	'meta' => 9, // NORMAL \| VOID_TAG
163	'meter' => 1,
164	'nav' => 17, // NORMAL \| AUTOCLOSE_P,
165	'noscript' => 65, // NORMAL \| BLOCK_TAG
166	'object' => 1,
167	'ol' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
168	'optgroup' => 1,
169	'option' => 1,
170	'output' => 65, // NORMAL \| BLOCK_TAG
171	'p' => 209, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG \| BLOCK_ONLY_INLINE
172	'param' => 9, // NORMAL \| VOID_TAG
173	'pre' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
174	'progress' => 1,
175	'q' => 1,
176	'rp' => 1,
177	'rt' => 1,
178	'ruby' => 1,
179	's' => 1,
180	'samp' => 1,
181	'script' => 3, // NORMAL \| TEXT_RAW
182	'section' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
183	'select' => 1,
184	'small' => 1,
185	'source' => 9, // NORMAL \| VOID_TAG
186	'span' => 1,
187	'strong' => 1,
188	'style' => 3, // NORMAL \| TEXT_RAW
189	'sub' => 1,
190	'summary' => 17, // NORMAL \| AUTOCLOSE_P,
191	'sup' => 1,
192	'table' => 65, // NORMAL \| BLOCK_TAG
193	'tbody' => 1,
194	'td' => 1,
195	'textarea' => 5, // NORMAL \| TEXT_RCDATA
196	'tfoot' => 65, // NORMAL \| BLOCK_TAG
197	'th' => 1,
198	'thead' => 1,
199	'time' => 1,
200	'title' => 5, // NORMAL \| TEXT_RCDATA
201	'tr' => 1,
202	'track' => 9, // NORMAL \| VOID_TAG
203	'u' => 1,
204	'ul' => 81, // NORMAL \| AUTOCLOSE_P \| BLOCK_TAG
205	'var' => 1,
206	'video' => 1,
207	'wbr' => 9, // NORMAL \| VOID_TAG
208
209	// Legacy?
210	'basefont' => 8, // VOID_TAG
211	'bgsound' => 8, // VOID_TAG
212	'noframes' => 2, // RAW_TEXT
213	'frame' => 9, // NORMAL \| VOID_TAG
214	'frameset' => 1,
215	'center' => 16,
216	'dir' => 16,
217	'listing' => 16, // AUTOCLOSE_P
218	'plaintext' => 48, // AUTOCLOSE_P \| TEXT_PLAINTEXT
219	'applet' => 0,
220	'marquee' => 0,
221	'isindex' => 8, // VOID_TAG
222	'xmp' => 20, // AUTOCLOSE_P \| VOID_TAG \| RAW_TEXT
223	'noembed' => 2, // RAW_TEXT
224	);
225
226	/**
227	* The MathML elements.
228	* See http://www.w3.org/wiki/MathML/Elements.
229	*
230	* In our case we are only concerned with presentation MathML and not content
231	* MathML. There is a nice list of this subset at https://developer.mozilla.org/en-US/docs/MathML/Element.
232	*
233	* @var array
234	*/
235	public static $mathml = array(
236	'maction' => 1,
237	'maligngroup' => 1,
238	'malignmark' => 1,
239	'math' => 1,
240	'menclose' => 1,
241	'merror' => 1,
242	'mfenced' => 1,
243	'mfrac' => 1,
244	'mglyph' => 1,
245	'mi' => 1,
246	'mlabeledtr' => 1,
247	'mlongdiv' => 1,
248	'mmultiscripts' => 1,
249	'mn' => 1,
250	'mo' => 1,
251	'mover' => 1,
252	'mpadded' => 1,
253	'mphantom' => 1,
254	'mroot' => 1,
255	'mrow' => 1,
256	'ms' => 1,
257	'mscarries' => 1,
258	'mscarry' => 1,
259	'msgroup' => 1,
260	'msline' => 1,
261	'mspace' => 1,
262	'msqrt' => 1,
263	'msrow' => 1,
264	'mstack' => 1,
265	'mstyle' => 1,
266	'msub' => 1,
267	'msup' => 1,
268	'msubsup' => 1,
269	'mtable' => 1,
270	'mtd' => 1,
271	'mtext' => 1,
272	'mtr' => 1,
273	'munder' => 1,
274	'munderover' => 1,
275	);
276
277	/**
278	* The svg elements.
279	*
280	* The Mozilla documentation has a good list at https://developer.mozilla.org/en-US/docs/SVG/Element.
281	* The w3c list appears to be lacking in some areas like filter effect elements.
282	* That list can be found at http://www.w3.org/wiki/SVG/Elements.
283	*
284	* Note, FireFox appears to do a better job rendering filter effects than chrome.
285	* While they are in the spec I'm not sure how widely implemented they are.
286	*
287	* @var array
288	*/
289	public static $svg = array(
290	'a' => 1,
291	'altGlyph' => 1,
292	'altGlyphDef' => 1,
293	'altGlyphItem' => 1,
294	'animate' => 1,
295	'animateColor' => 1,
296	'animateMotion' => 1,
297	'animateTransform' => 1,
298	'circle' => 1,
299	'clipPath' => 1,
300	'color-profile' => 1,
301	'cursor' => 1,
302	'defs' => 1,
303	'desc' => 1,
304	'ellipse' => 1,
305	'feBlend' => 1,
306	'feColorMatrix' => 1,
307	'feComponentTransfer' => 1,
308	'feComposite' => 1,
309	'feConvolveMatrix' => 1,
310	'feDiffuseLighting' => 1,
311	'feDisplacementMap' => 1,
312	'feDistantLight' => 1,
313	'feDropShadow' => 1,
314	'feFlood' => 1,
315	'feFuncA' => 1,
316	'feFuncB' => 1,
317	'feFuncG' => 1,
318	'feFuncR' => 1,
319	'feGaussianBlur' => 1,
320	'feImage' => 1,
321	'feMerge' => 1,
322	'feMergeNode' => 1,
323	'feMorphology' => 1,
324	'feOffset' => 1,
325	'fePointLight' => 1,
326	'feSpecularLighting' => 1,
327	'feSpotLight' => 1,
328	'feTile' => 1,
329	'feTurbulence' => 1,
330	'filter' => 1,
331	'font' => 1,
332	'font-face' => 1,
333	'font-face-format' => 1,
334	'font-face-name' => 1,
335	'font-face-src' => 1,
336	'font-face-uri' => 1,
337	'foreignObject' => 1,
338	'g' => 1,
339	'glyph' => 1,
340	'glyphRef' => 1,
341	'hkern' => 1,
342	'image' => 1,
343	'line' => 1,
344	'linearGradient' => 1,
345	'marker' => 1,
346	'mask' => 1,
347	'metadata' => 1,
348	'missing-glyph' => 1,
349	'mpath' => 1,
350	'path' => 1,
351	'pattern' => 1,
352	'polygon' => 1,
353	'polyline' => 1,
354	'radialGradient' => 1,
355	'rect' => 1,
356	'script' => 3, // NORMAL \| RAW_TEXT
357	'set' => 1,
358	'stop' => 1,
359	'style' => 3, // NORMAL \| RAW_TEXT
360	'svg' => 1,
361	'switch' => 1,
362	'symbol' => 1,
363	'text' => 1,
364	'textPath' => 1,
365	'title' => 1,
366	'tref' => 1,
367	'tspan' => 1,
368	'use' => 1,
369	'view' => 1,
370	'vkern' => 1,
371	);
372
373	/**
374	* Some attributes in SVG are case sensitive.
375	*
376	* This map contains key/value pairs with the key as the lowercase attribute
377	* name and the value with the correct casing.
378	*/
379	public static $svgCaseSensitiveAttributeMap = array(
380	'attributename' => 'attributeName',
381	'attributetype' => 'attributeType',
382	'basefrequency' => 'baseFrequency',
383	'baseprofile' => 'baseProfile',
384	'calcmode' => 'calcMode',
385	'clippathunits' => 'clipPathUnits',
386	'contentscripttype' => 'contentScriptType',
387	'contentstyletype' => 'contentStyleType',
388	'diffuseconstant' => 'diffuseConstant',
389	'edgemode' => 'edgeMode',
390	'externalresourcesrequired' => 'externalResourcesRequired',
391	'filterres' => 'filterRes',
392	'filterunits' => 'filterUnits',
393	'glyphref' => 'glyphRef',
394	'gradienttransform' => 'gradientTransform',
395	'gradientunits' => 'gradientUnits',
396	'kernelmatrix' => 'kernelMatrix',
397	'kernelunitlength' => 'kernelUnitLength',
398	'keypoints' => 'keyPoints',
399	'keysplines' => 'keySplines',
400	'keytimes' => 'keyTimes',
401	'lengthadjust' => 'lengthAdjust',
402	'limitingconeangle' => 'limitingConeAngle',
403	'markerheight' => 'markerHeight',
404	'markerunits' => 'markerUnits',
405	'markerwidth' => 'markerWidth',
406	'maskcontentunits' => 'maskContentUnits',
407	'maskunits' => 'maskUnits',
408	'numoctaves' => 'numOctaves',
409	'pathlength' => 'pathLength',
410	'patterncontentunits' => 'patternContentUnits',
411	'patterntransform' => 'patternTransform',
412	'patternunits' => 'patternUnits',
413	'pointsatx' => 'pointsAtX',
414	'pointsaty' => 'pointsAtY',
415	'pointsatz' => 'pointsAtZ',
416	'preservealpha' => 'preserveAlpha',
417	'preserveaspectratio' => 'preserveAspectRatio',
418	'primitiveunits' => 'primitiveUnits',
419	'refx' => 'refX',
420	'refy' => 'refY',
421	'repeatcount' => 'repeatCount',
422	'repeatdur' => 'repeatDur',
423	'requiredextensions' => 'requiredExtensions',
424	'requiredfeatures' => 'requiredFeatures',
425	'specularconstant' => 'specularConstant',
426	'specularexponent' => 'specularExponent',
427	'spreadmethod' => 'spreadMethod',
428	'startoffset' => 'startOffset',
429	'stddeviation' => 'stdDeviation',
430	'stitchtiles' => 'stitchTiles',
431	'surfacescale' => 'surfaceScale',
432	'systemlanguage' => 'systemLanguage',
433	'tablevalues' => 'tableValues',
434	'targetx' => 'targetX',
435	'targety' => 'targetY',
436	'textlength' => 'textLength',
437	'viewbox' => 'viewBox',
438	'viewtarget' => 'viewTarget',
439	'xchannelselector' => 'xChannelSelector',
440	'ychannelselector' => 'yChannelSelector',
441	'zoomandpan' => 'zoomAndPan',
442	);
443
444	/**
445	* Some SVG elements are case sensitive.
446	* This map contains these.
447	*
448	* The map contains key/value store of the name is lowercase as the keys and
449	* the correct casing as the value.
450	*/
451	public static $svgCaseSensitiveElementMap = array(
452	'altglyph' => 'altGlyph',
453	'altglyphdef' => 'altGlyphDef',
454	'altglyphitem' => 'altGlyphItem',
455	'animatecolor' => 'animateColor',
456	'animatemotion' => 'animateMotion',
457	'animatetransform' => 'animateTransform',
458	'clippath' => 'clipPath',
459	'feblend' => 'feBlend',
460	'fecolormatrix' => 'feColorMatrix',
461	'fecomponenttransfer' => 'feComponentTransfer',
462	'fecomposite' => 'feComposite',
463	'feconvolvematrix' => 'feConvolveMatrix',
464	'fediffuselighting' => 'feDiffuseLighting',
465	'fedisplacementmap' => 'feDisplacementMap',
466	'fedistantlight' => 'feDistantLight',
467	'fedropshadow' => 'feDropShadow',
468	'feflood' => 'feFlood',
469	'fefunca' => 'feFuncA',
470	'fefuncb' => 'feFuncB',
471	'fefuncg' => 'feFuncG',
472	'fefuncr' => 'feFuncR',
473	'fegaussianblur' => 'feGaussianBlur',
474	'feimage' => 'feImage',
475	'femerge' => 'feMerge',
476	'femergenode' => 'feMergeNode',
477	'femorphology' => 'feMorphology',
478	'feoffset' => 'feOffset',
479	'fepointlight' => 'fePointLight',
480	'fespecularlighting' => 'feSpecularLighting',
481	'fespotlight' => 'feSpotLight',
482	'fetile' => 'feTile',
483	'feturbulence' => 'feTurbulence',
484	'foreignobject' => 'foreignObject',
485	'glyphref' => 'glyphRef',
486	'lineargradient' => 'linearGradient',
487	'radialgradient' => 'radialGradient',
488	'textpath' => 'textPath',
489	);
490
491	/**
492	* Check whether the given element meets the given criterion.
493	*
494	* Example:
495	*
496	* Elements::isA('script', Elements::TEXT_RAW); // Returns true.
497	*
498	* Elements::isA('script', Elements::TEXT_RCDATA); // Returns false.
499	*
500	* @param string $name The element name.
501	* @param int $mask One of the constants on this class.
502	*
503	* @return bool true if the element matches the mask, false otherwise.
504	*/
505	public static function isA($name, $mask)
506	{
507	return (static::element($name) & $mask) === $mask;
508	}
509
510	/**
511	* Test if an element is a valid html5 element.
512	*
513	* @param string $name The name of the element.
514	*
515	* @return bool true if a html5 element and false otherwise.
516	*/
517	public static function isHtml5Element($name)
518	{
519	// html5 element names are case insensitive. Forcing lowercase for the check.
520	// Do we need this check or will all data passed here already be lowercase?
521	return isset(static::$html5[strtolower($name)]);
522	}
523
524	/**
525	* Test if an element name is a valid MathML presentation element.
526	*
527	* @param string $name The name of the element.
528	*
529	* @return bool true if a MathML name and false otherwise.
530	*/
531	public static function isMathMLElement($name)
532	{
533	// MathML is case-sensitive unlike html5 elements.
534	return isset(static::$mathml[$name]);
535	}
536
537	/**
538	* Test if an element is a valid SVG element.
539	*
540	* @param string $name The name of the element.
541	*
542	* @return bool true if a SVG element and false otherise.
543	*/
544	public static function isSvgElement($name)
545	{
546	// SVG is case-sensitive unlike html5 elements.
547	return isset(static::$svg[$name]);
548	}
549
550	/**
551	* Is an element name valid in an html5 document.
552	* This includes html5 elements along with other allowed embedded content
553	* such as svg and mathml.
554	*
555	* @param string $name The name of the element.
556	*
557	* @return bool true if valid and false otherwise.
558	*/
559	public static function isElement($name)
560	{
561	return static::isHtml5Element($name) \|\| static::isMathMLElement($name) \|\| static::isSvgElement($name);
562	}
563
564	/**
565	* Get the element mask for the given element name.
566	*
567	* @param string $name The name of the element.
568	*
569	* @return int the element mask.
570	*/
571	public static function element($name)
572	{
573	if (isset(static::$html5[$name])) {
574	return static::$html5[$name];
575	}
576	if (isset(static::$svg[$name])) {
577	return static::$svg[$name];
578	}
579	if (isset(static::$mathml[$name])) {
580	return static::$mathml[$name];
581	}
582
583	return 0;
584	}
585
586	/**
587	* Normalize a SVG element name to its proper case and form.
588	*
589	* @param string $name The name of the element.
590	*
591	* @return string the normalized form of the element name.
592	*/
593	public static function normalizeSvgElement($name)
594	{
595	$name = strtolower($name);
596	if (isset(static::$svgCaseSensitiveElementMap[$name])) {
597	$name = static::$svgCaseSensitiveElementMap[$name];
598	}
599
600	return $name;
601	}
602
603	/**
604	* Normalize a SVG attribute name to its proper case and form.
605	*
606	* @param string $name The name of the attribute.
607	*
608	* @return string The normalized form of the attribute name.
609	*/
610	public static function normalizeSvgAttribute($name)
611	{
612	$name = strtolower($name);
613	if (isset(static::$svgCaseSensitiveAttributeMap[$name])) {
614	$name = static::$svgCaseSensitiveAttributeMap[$name];
615	}
616
617	return $name;
618	}
619
620	/**
621	* Normalize a MathML attribute name to its proper case and form.
622	* Note, all MathML element names are lowercase.
623	*
624	* @param string $name The name of the attribute.
625	*
626	* @return string The normalized form of the attribute name.
627	*/
628	public static function normalizeMathMlAttribute($name)
629	{
630	$name = strtolower($name);
631
632	// Only one attribute has a mixed case form for MathML.
633	if ('definitionurl' === $name) {
634	$name = 'definitionURL';
635	}
636
637	return $name;
638	}
639	}
640