Normalizer::getOutputContext() - Code Metrics - Inspection of "WiP" - s9e/TextFormatter - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Branch — TemplateParserRefactor (06d644)

by Josh

created 2018-02-15 14:09 UTC

Normalizer::getOutputContext() A

↳ Parent: Normalizer

Complexity

Conditions	3
Paths	3

Size

Total Lines	17
Code Lines	9

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
dl	0
loc	17
rs	9.4285
c	1
b	0
f	0
cc	3
eloc	9
nc	3
nop	1

<?php

/**
* @package   s9e\TextFormatter
* @copyright Copyright (c) 2010-2018 The s9e Authors
* @license   http://www.opensource.org/licenses/mit-license.php The MIT License
*/
namespace s9e\TextFormatter\Configurator\Helpers\TemplateParser;

use DOMDocument;
use DOMElement;
use DOMNode;
use s9e\TextFormatter\Configurator\Helpers\XPathHelper;

class Normalizer extends IRProcessor
{
	/**
	* @var Optimizer
	*/
	protected $optimizer;

	/**
	* @var string Regexp that matches the names of all void elements
	* @link http://www.w3.org/TR/html-markup/syntax.html#void-elements
	*/
	public $voidRegexp = '/^(?:area|base|br|col|command|embed|hr|img|input|keygen|link|meta|param|source|track|wbr)$/Di';

	/**
	* @param  Optimizer $optimizer
	* @return void
	*/
	public function __construct(Optimizer $optimizer)
	{
		$this->optimizer = $optimizer;
	}

	/**
	* Normalize an IR
	*
	* @param  DOMDocument $ir
	* @return void
	*/
	public function normalize(DOMDocument $ir)
	{
		$this->createXPath($ir);
		$this->addDefaultCase($ir);
		$this->addElementIds($ir);
		$this->addCloseTagElements($ir);
		$this->markEmptyElements($ir);
		$this->optimizer->optimize($ir);
		$this->markConditionalCloseTagElements($ir);
		$this->setOutputContext($ir);
		$this->markBranchTables($ir);
	}

	/**
	* Add <closeTag/> elements everywhere an open start tag should be closed
	*
	* @param  DOMDocument $ir
	* @return void
	*/
	protected function addCloseTagElements(DOMDocument $ir)
	{
		$exprs = [
			'//applyTemplates[not(ancestor::attribute)]',
			'//comment',
			'//element',
			'//output[not(ancestor::attribute)]'
		];
		foreach ($this->query(implode('|', $exprs)) as $node)
		{
			$parentElementId = $this->getParentElementId($node);
			if (isset($parentElementId))
			{
				$node->parentNode
				     ->insertBefore($ir->createElement('closeTag'), $node)
				     ->setAttribute('id', $parentElementId);
			}

			// Append a <closeTag/> to <element/> nodes to ensure that empty elements get closed
			if ($node->nodeName === 'element')
			{
				$id = $node->getAttribute('id');
				$this->appendElement($node, 'closeTag')->setAttribute('id', $id);
			}
		}
	}

	/**
	* Add an empty default <case/> to <switch/> nodes that don't have one
	*
	* @param  DOMDocument $ir
	* @return void
	*/
	protected function addDefaultCase(DOMDocument $ir)
	{
		foreach ($this->query('//switch[not(case[not(@test)])]') as $switch)
		{
			$this->appendElement($switch, 'case');
		}
	}

	/**
	* Add an id attribute to <element/> nodes
	*
	* @param  DOMDocument $ir
	* @return void
	*/
	protected function addElementIds(DOMDocument $ir)
	{
		$id = 0;
		foreach ($ir->getElementsByTagName('element') as $element)
		{
			$element->setAttribute('id', ++$id);
		}
	}

	/**
	* Get the context type for given output element
	*
	* @param  DOMNode $output
	* @return string
	*/
	protected function getOutputContext(DOMNode $output)
	{
		$contexts = [
			'boolean(ancestor::attribute)'             => 'attribute',
			'@disable-output-escaping="yes"'           => 'raw',
			'count(ancestor::element[@name="script"])' => 'raw'
		];
		foreach ($contexts as $expr => $context)
		{
			if ($this->evaluate($expr, $output))
			{
				return $context;
			}
		}

		return 'text';
	}

	/**
	* Get the ID of the closest "element" ancestor
	*
	* @param  DOMNode     $node Context node
	* @return string|null
	*/
	protected function getParentElementId(DOMNode $node)
	{
		$parentNode = $node->parentNode;
		while (isset($parentNode))
		{
			if ($parentNode->nodeName === 'element')
			{
				return $parentNode->getAttribute('id');
			}
			$parentNode = $parentNode->parentNode;
		}
	}

	/**
	* Test whether given element will be empty at runtime (no content, no children)
	*
	* @param  DOMElement $ir Element in the IR
	* @return string         'yes', 'maybe' or 'no'
	*/
	protected function isEmpty(DOMElement $ir)
	{
		// Comments and elements count as not-empty and literal output is sure to output something
		if ($this->evaluate('count(comment | element | output[@type="literal"])', $ir))
		{
			return 'no';
		}

		// Test all branches of a <switch/>
		// NOTE: this assumes that <switch/> are normalized to always have a default <case/>
		$cases = [];
		foreach ($this->query('switch/case', $ir) as $case)
		{
			$cases[$this->isEmpty($case)] = 1;
		}

		if (isset($cases['maybe']))
		{
			return 'maybe';
		}

		if (isset($cases['no']))
		{
			// If all the cases are not-empty, the element is not-empty
			if (!isset($cases['yes']))
			{
				return 'no';
			}

			// Some 'yes' and some 'no', the element is a 'maybe'
			return 'maybe';
		}

		// Test for <apply-templates/> or XPath output
		if ($this->evaluate('count(applyTemplates | output[@type="xpath"])', $ir))
		{
			// We can't know in advance whether those will produce output
			return 'maybe';
		}

		return 'yes';
	}

	/**
	* Mark switch elements that are used as branch tables
	*
	* If a switch is used for a series of equality tests against the same attribute or variable, the
	* attribute/variable is stored within the switch as "branch-key" and the values it is compared
	* against are stored JSON-encoded in the case as "branch-values". It can be used to create
	* optimized branch tables
	*
	* @param  DOMDocument $ir
	* @return void
	*/
	protected function markBranchTables(DOMDocument $ir)
	{

		// Iterate over switch elements that have at least two case children with a test attribute
		foreach ($this->query('//switch[case[2][@test]]') as $switch)
		{
			$key = null;
			$branchValues = [];

			foreach ($switch->childNodes as $i => $case)
			{
				if (!$case->hasAttribute('test'))
				{
					continue;
				}

				$map = XPathHelper::parseEqualityExpr($case->getAttribute('test'));

				// Test whether the expression matches an equality
				if ($map === false)
				{
					continue 2;
				}

				// Abort if there's more than 1 variable used
				if (count($map) !== 1)
				{
					continue 2;
				}

				// Test whether it uses the same key
				if (isset($key) && $key !== key($map))
				{
					continue 2;
				}

				$key = key($map);
				$branchValues[$i] = end($map);
			}

			$switch->setAttribute('branch-key', $key);
			foreach ($branchValues as $i => $values)
			{
				sort($values);
				$switch->childNodes->item($i)->setAttribute('branch-values', serialize($values));
			}
		}
	}

	/**
	* Mark conditional <closeTag/> nodes
	*
	* @param  DOMDocument $ir
	* @return void
	*/
	protected function markConditionalCloseTagElements(DOMDocument $ir)
	{
		foreach ($ir->getElementsByTagName('closeTag') as $closeTag)
		{
			$id = $closeTag->getAttribute('id');

			// For each <switch/> ancestor, look for a <closeTag/> and that is either a sibling or
			// the descendant of a sibling, and that matches the id
			$query = 'ancestor::switch/'
			       . 'following-sibling::*/'
			       . 'descendant-or-self::closeTag[@id = "' . $id . '"]';
			foreach ($this->query($query, $closeTag) as $following)
			{
				// Mark following <closeTag/> nodes to indicate that the status of this tag must
				// be checked before it is closed
				$following->setAttribute('check', '');

				// Mark the current <closeTag/> to indicate that it must set a flag to indicate
				// that its tag has been closed
				$closeTag->setAttribute('set', '');
			}
		}
	}

	/**
	* Mark void elements and elements with no content
	*
	* @param  DOMDocument $ir
	* @return void
	*/
	protected function markEmptyElements(DOMDocument $ir)
	{
		foreach ($ir->getElementsByTagName('element') as $element)
		{
			// Test whether this element is (maybe) void
			$elName = $element->getAttribute('name');
			if (strpos($elName, '{') !== false)
			{
				// Dynamic element names must be checked at runtime
				$element->setAttribute('void', 'maybe');
			}
			elseif (preg_match($this->voidRegexp, $elName))
			{
				// Static element names can be checked right now
				$element->setAttribute('void', 'yes');
			}

			// Find whether this element is empty
			$isEmpty = $this->isEmpty($element);
			if ($isEmpty === 'yes' || $isEmpty === 'maybe')
			{
				$element->setAttribute('empty', $isEmpty);
			}
		}
	}

	/**
	* Fill in output context
	*
	* @param  DOMDocument $ir
	* @return void
	*/
	protected function setOutputContext(DOMDocument $ir)
	{
		foreach ($ir->getElementsByTagName('output') as $output)
		{
			$output->setAttribute('escape', $this->getOutputContext($output));
		}
	}
}

1			<?php
2
3			/**
4			* @package s9e\TextFormatter
5			* @copyright Copyright (c) 2010-2018 The s9e Authors
6			* @license http://www.opensource.org/licenses/mit-license.php The MIT License
7			*/
8			namespace s9e\TextFormatter\Configurator\Helpers\TemplateParser;
9
10			use DOMDocument;
11			use DOMElement;
12			use DOMNode;
13			use s9e\TextFormatter\Configurator\Helpers\XPathHelper;
14
15			class Normalizer extends IRProcessor
16			{
17			/**
18			* @var Optimizer
19			*/
20			protected $optimizer;
21
22			/**
23			* @var string Regexp that matches the names of all void elements
24			* @link http://www.w3.org/TR/html-markup/syntax.html#void-elements
25			*/
26			public $voidRegexp = '/^(?:area\|base\|br\|col\|command\|embed\|hr\|img\|input\|keygen\|link\|meta\|param\|source\|track\|wbr)$/Di';
27
28			/**
29			* @param Optimizer $optimizer
30			* @return void
31			*/
32			public function __construct(Optimizer $optimizer)
33			{
34			$this->optimizer = $optimizer;
35			}
36
37			/**
38			* Normalize an IR
39			*
40			* @param DOMDocument $ir
41			* @return void
42			*/
43			public function normalize(DOMDocument $ir)
44			{
45			$this->createXPath($ir);
46			$this->addDefaultCase($ir);
47			$this->addElementIds($ir);
48			$this->addCloseTagElements($ir);
49			$this->markEmptyElements($ir);
50			$this->optimizer->optimize($ir);
51			$this->markConditionalCloseTagElements($ir);
52			$this->setOutputContext($ir);
53			$this->markBranchTables($ir);
54			}
55
56			/**
57			* Add <closeTag/> elements everywhere an open start tag should be closed
58			*
59			* @param DOMDocument $ir
60			* @return void
61			*/
62			protected function addCloseTagElements(DOMDocument $ir)
63			{
64			$exprs = [
65			'//applyTemplates[not(ancestor::attribute)]',
66			'//comment',
67			'//element',
68			'//output[not(ancestor::attribute)]'
69			];
70			foreach ($this->query(implode('\|', $exprs)) as $node)
71			{
72			$parentElementId = $this->getParentElementId($node);
73			if (isset($parentElementId))
74			{
75			$node->parentNode
76			->insertBefore($ir->createElement('closeTag'), $node)
77			->setAttribute('id', $parentElementId);
78			}
79
80			// Append a <closeTag/> to <element/> nodes to ensure that empty elements get closed
81			if ($node->nodeName === 'element')
82			{
83			$id = $node->getAttribute('id');
84			$this->appendElement($node, 'closeTag')->setAttribute('id', $id);
85			}
86			}
87			}
88
89			/**
90			* Add an empty default <case/> to <switch/> nodes that don't have one
91			*
92			* @param DOMDocument $ir
93			* @return void
94			*/
95			protected function addDefaultCase(DOMDocument $ir)
96			{
97			foreach ($this->query('//switch[not(case[not(@test)])]') as $switch)
98			{
99			$this->appendElement($switch, 'case');
100			}
101			}
102
103			/**
104			* Add an id attribute to <element/> nodes
105			*
106			* @param DOMDocument $ir
107			* @return void
108			*/
109			protected function addElementIds(DOMDocument $ir)
110			{
111			$id = 0;
112			foreach ($ir->getElementsByTagName('element') as $element)
113			{
114			$element->setAttribute('id', ++$id);
115			}
116			}
117
118			/**
119			* Get the context type for given output element
120			*
121			* @param DOMNode $output
122			* @return string
123			*/
124			protected function getOutputContext(DOMNode $output)
125			{
126			$contexts = [
127			'boolean(ancestor::attribute)' => 'attribute',
128			'@disable-output-escaping="yes"' => 'raw',
129			'count(ancestor::element[@name="script"])' => 'raw'
130			];
131			foreach ($contexts as $expr => $context)
132			{
133			if ($this->evaluate($expr, $output))
134			{
135			return $context;
136			}
137			}
138
139			return 'text';
140			}
141
142			/**
143			* Get the ID of the closest "element" ancestor
144			*
145			* @param DOMNode $node Context node
146			* @return string\|null
147			*/
148			protected function getParentElementId(DOMNode $node)
149			{
150			$parentNode = $node->parentNode;
151			while (isset($parentNode))
152			{
153			if ($parentNode->nodeName === 'element')
154			{
155			return $parentNode->getAttribute('id');
156			}
157			$parentNode = $parentNode->parentNode;
158			}
159			}
160
161			/**
162			* Test whether given element will be empty at runtime (no content, no children)
163			*
164			* @param DOMElement $ir Element in the IR
165			* @return string 'yes', 'maybe' or 'no'
166			*/
167			protected function isEmpty(DOMElement $ir)
168			{
169			// Comments and elements count as not-empty and literal output is sure to output something
170			if ($this->evaluate('count(comment \| element \| output[@type="literal"])', $ir))
171			{
172			return 'no';
173			}
174
175			// Test all branches of a <switch/>
176			// NOTE: this assumes that <switch/> are normalized to always have a default <case/>
177			$cases = [];
178			foreach ($this->query('switch/case', $ir) as $case)
179			{
180			$cases[$this->isEmpty($case)] = 1;
181			}
182
183			if (isset($cases['maybe']))
184			{
185			return 'maybe';
186			}
187
188			if (isset($cases['no']))
189			{
190			// If all the cases are not-empty, the element is not-empty
191			if (!isset($cases['yes']))
192			{
193			return 'no';
194			}
195
196			// Some 'yes' and some 'no', the element is a 'maybe'
197			return 'maybe';
198			}
199
200			// Test for <apply-templates/> or XPath output
201			if ($this->evaluate('count(applyTemplates \| output[@type="xpath"])', $ir))
202			{
203			// We can't know in advance whether those will produce output
204			return 'maybe';
205			}
206
207			return 'yes';
208			}
209
210			/**
211			* Mark switch elements that are used as branch tables
212			*
213			* If a switch is used for a series of equality tests against the same attribute or variable, the
214			* attribute/variable is stored within the switch as "branch-key" and the values it is compared
215			* against are stored JSON-encoded in the case as "branch-values". It can be used to create
216			* optimized branch tables
217			*
218			* @param DOMDocument $ir
219			* @return void
220			*/
221			protected function markBranchTables(DOMDocument $ir)
222			{
223
224			// Iterate over switch elements that have at least two case children with a test attribute
225			foreach ($this->query('//switch[case[2][@test]]') as $switch)
226			{
227			$key = null;
228			$branchValues = [];
229
230			foreach ($switch->childNodes as $i => $case)
231			{
232			if (!$case->hasAttribute('test'))
233			{
234			continue;
235			}
236
237			$map = XPathHelper::parseEqualityExpr($case->getAttribute('test'));
238
239			// Test whether the expression matches an equality
240			if ($map === false)
241			{
242			continue 2;
243			}
244
245			// Abort if there's more than 1 variable used
246			if (count($map) !== 1)
247			{
248			continue 2;
249			}
250
251			// Test whether it uses the same key
252			if (isset($key) && $key !== key($map))
253			{
254			continue 2;
255			}
256
257			$key = key($map);
258			$branchValues[$i] = end($map);
259			}
260
261			$switch->setAttribute('branch-key', $key);
262			foreach ($branchValues as $i => $values)
263			{
264			sort($values);
265			$switch->childNodes->item($i)->setAttribute('branch-values', serialize($values));
266			}
267			}
268			}
269
270			/**
271			* Mark conditional <closeTag/> nodes
272			*
273			* @param DOMDocument $ir
274			* @return void
275			*/
276			protected function markConditionalCloseTagElements(DOMDocument $ir)
277			{
278			foreach ($ir->getElementsByTagName('closeTag') as $closeTag)
279			{
280			$id = $closeTag->getAttribute('id');
281
282			// For each <switch/> ancestor, look for a <closeTag/> and that is either a sibling or
283			// the descendant of a sibling, and that matches the id
284			$query = 'ancestor::switch/'
285			. 'following-sibling::*/'
286			. 'descendant-or-self::closeTag[@id = "' . $id . '"]';
287			foreach ($this->query($query, $closeTag) as $following)
288			{
289			// Mark following <closeTag/> nodes to indicate that the status of this tag must
290			// be checked before it is closed
291			$following->setAttribute('check', '');
292
293			// Mark the current <closeTag/> to indicate that it must set a flag to indicate
294			// that its tag has been closed
295			$closeTag->setAttribute('set', '');
296			}
297			}
298			}
299
300			/**
301			* Mark void elements and elements with no content
302			*
303			* @param DOMDocument $ir
304			* @return void
305			*/
306			protected function markEmptyElements(DOMDocument $ir)
307			{
308			foreach ($ir->getElementsByTagName('element') as $element)
309			{
310			// Test whether this element is (maybe) void
311			$elName = $element->getAttribute('name');
312			if (strpos($elName, '{') !== false)
313			{
314			// Dynamic element names must be checked at runtime
315			$element->setAttribute('void', 'maybe');
316			}
317			elseif (preg_match($this->voidRegexp, $elName))
318			{
319			// Static element names can be checked right now
320			$element->setAttribute('void', 'yes');
321			}
322
323			// Find whether this element is empty
324			$isEmpty = $this->isEmpty($element);
325			if ($isEmpty === 'yes' \|\| $isEmpty === 'maybe')
326			{
327			$element->setAttribute('empty', $isEmpty);
328			}
329			}
330			}
331
332			/**
333			* Fill in output context
334			*
335			* @param DOMDocument $ir
336			* @return void
337			*/
338			protected function setOutputContext(DOMDocument $ir)
339			{
340			foreach ($ir->getElementsByTagName('output') as $output)
341			{
342			$output->setAttribute('escape', $this->getOutputContext($output));
343			}
344			}
345			}

s9e / TextFormatter

Branch — TemplateParserRefactor (06d644)

Normalizer::getOutputContext() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like