Configurator::getSiteHelper() - Code Metrics - Inspection of "MediaEmbed: added SiteHelpers API" - s9e/TextFormatter - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( a025c6...613f63 )

by Josh

created 2024-07-24 13:11 UTC

Configurator::getSiteHelper() A

↳ Parent: Configurator

Complexity

Conditions	1
Paths	1

Size

Total Lines	5
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	1
CRAP Score	1

Importance

Changes

Metric	Value
eloc	2
c	0
b	0
f	0
dl	0
loc	5
rs	10
ccs	1
cts	1
cp	1
cc	1
nc	1
nop	1
crap	1

<?php

/**
* @package   s9e\TextFormatter
* @copyright Copyright (c) The s9e authors
* @license   http://www.opensource.org/licenses/mit-license.php The MIT License
*/
namespace s9e\TextFormatter\Plugins\MediaEmbed;

use InvalidArgumentException;
use RuntimeException;
use s9e\TextFormatter\Configurator\Helpers\FilterHelper;
use s9e\TextFormatter\Configurator\Items\Regexp;
use s9e\TextFormatter\Configurator\Items\Tag;
use s9e\TextFormatter\Configurator\JavaScript\Dictionary;
use s9e\TextFormatter\Plugins\ConfiguratorBase;
use s9e\TextFormatter\Plugins\MediaEmbed\Configurator\Collections\CachedDefinitionCollection;
use s9e\TextFormatter\Plugins\MediaEmbed\Configurator\SiteHelpers\AbstractSiteHelper;
use s9e\TextFormatter\Plugins\MediaEmbed\Configurator\TemplateBuilder;

class Configurator extends ConfiguratorBase
{
	/**
	* @var array List of filters that are explicitly allowed in attribute definitions
	*/
	public $allowedFilters = ['htmlspecialchars_decode', 'stripslashes', 'str_replace', 'urldecode'];

	/**
	* @var bool Whether to create the MEDIA BBCode
	*/
	protected $createMediaBBCode = true;

	/**
	* @var Configurator\Collections\SiteDefinitionCollection Default sites
	*/
	public $defaultSites;

	/**
	* {@inheritdoc}
	*/
	protected $quickMatch = '://';

	/**
	* {@inheritdoc}
	*/
	protected $regexp = '/\\bhttps?:\\/\\/[^["\'\\s]+/Si';

	/**
	* @var array Configured sites
	*/
	protected $sites = [];

	/**
	* @var string Name of the tag used to handle embeddable URLs
	*/
	protected $tagName = 'MEDIA';

	/**
	* @var TemplateBuilder
	*/
	protected $templateBuilder;

	/**
	* {@inheritdoc}
	*/
	protected function setUp()
	{
		$this->defaultSites    = new CachedDefinitionCollection;
		$this->templateBuilder = new TemplateBuilder;

		$this->configurator->registeredVars['MediaEmbed.hosts'] = new Dictionary;
		$this->configurator->registeredVars['MediaEmbed.sites'] = new Dictionary;

		// Create a MEDIA tag
		$this->createMediaTag();

		// Create a [MEDIA] BBCode if applicable
		if ($this->createMediaBBCode)
		{
			$this->configurator->BBCodes->set($this->tagName, ['contentAttributes' => ['url']]);
		}
	}

	/**
	* {@inheritdoc}
	*/
	public function asConfig()
	{
		if (empty($this->sites))
		{
			return;
		}

		return [
			'quickMatch' => $this->quickMatch,
			'regexp'     => $this->regexp,
			'tagName'    => $this->tagName
		];
	}

	/**
	* Add a media site
	*
	* @param  string $siteId     Site's ID
	* @param  array  $siteConfig Site's config
	* @return Tag                Tag created for this site
	*/
	public function add($siteId, ?array $siteConfig = null)
	{
		// Normalize or retrieve the site definition
		$siteId = $this->normalizeId($siteId);
		if (isset($siteConfig))
		{
			$siteConfig = $this->defaultSites->normalizeValue($siteConfig);
		}
		else
		{
			$siteConfig = $this->defaultSites->get($siteId);
		}
		$siteConfig['extract'] = $this->convertRegexps($siteConfig['extract']);
		$siteConfig['scrape']  = $this->convertScrapes($siteConfig['scrape']);

		// Check the safety of attribute filters
		$this->checkAttributeFilters($siteConfig['attributes']);

		// Create the tag for this site
		$tag = $this->addTag($siteId, $siteConfig);

		// Update the configurator's data
		$this->sites[$siteId] = $siteConfig;
		foreach ($siteConfig['host'] as $host)
		{
			$this->configurator->registeredVars['MediaEmbed.hosts'][$host] = $siteId;
		}
		$this->configurator->registeredVars['MediaEmbed.sites'][$siteId] = [$siteConfig['extract'], $siteConfig['scrape']];

		return $tag;
	}

	public function getSiteHelper(string $siteId): AbstractSiteHelper
	{
		$className = $this->defaultSites->get($siteId)['helper'];

		return new $className($this->configurator);
	}

	/**
	* Return the list of configured sites
	*
	* @return array Site's ID as keys, site's config as values
	*/
	public function getSites()
	{
		return $this->sites;
	}

	/**
	* Create and return a tag that handles given media site
	*
	* @param  string $siteId
	* @param  array  $siteConfig
	* @return Tag
	*/
	protected function addTag($siteId, array $siteConfig)
	{
		$tag = new Tag([
			'attributes' => $this->getAttributesConfig($siteConfig),
			'rules'      => [
				'allowChild' => 'URL',
				'autoClose'  => true,
				'denyChild'  => [$siteId, $this->tagName]
			],
			'template'   => $this->templateBuilder->build($siteId, $siteConfig)
		]);

		$this->configurator->templateNormalizer->normalizeTag($tag);
		$this->configurator->templateChecker->checkTag($tag);
		$this->configurator->tags->add($siteId, $tag);

		return $tag;
	}

	/**
	* Check the safety of given attributes
	*
	* @param  array $attributes
	* @return void
	*/
	protected function checkAttributeFilters(array $attributes)
	{
		foreach ($attributes as $attrConfig)
		{
			if (empty($attrConfig['filterChain']))
			{
				continue;
			}
			foreach ($attrConfig['filterChain'] as $filter)
			{
				if (!FilterHelper::isAllowed($filter, $this->allowedFilters))
				{
					throw new RuntimeException("Filter '$filter' is not allowed in media sites");
				}
			}
		}
	}

	/**
	* Convert given regexp to a [regexp, map] pair
	*
	* @param  string $regexp Original regexp
	* @return array          [regexp, [list of captures' names]]
	*/
	protected function convertRegexp($regexp)
	{
		$regexp = new Regexp($regexp);

		return [$regexp, $regexp->getCaptureNames()];
	}

	/**
	* Convert a list of regexps
	*
	* @param  string[] $regexps Original list
	* @return array[]           Converted list
	*/
	protected function convertRegexps(array $regexps)
	{
		return array_map($this->convertRegexp(...), $regexps);
	}

	/**
	* Convert all regexps in a scraping config
	*
	* @param  array $config Original config
	* @return array         Converted config
	*/
	protected function convertScrapeConfig(array $config)
	{
		$config['extract'] = $this->convertRegexps($config['extract']);
		$config['match']   = $this->convertRegexps($config['match']);

		return $config;
	}

	/**
	* Convert all regexps in a list of scraping configs
	*
	* @param  array[] $scrapes Original config
	* @return array[]          Converted config
	*/
	protected function convertScrapes(array $scrapes)
	{
		return array_map($this->convertScrapeConfig(...), $scrapes);
	}

	/**
	* Create the default MEDIA tag
	*
	* @return void
	*/
	protected function createMediaTag()
	{
		$tag = $this->configurator->tags->add($this->tagName);

		// This tag should not need to be closed and should not contain itself
		$tag->rules->autoClose();
		$tag->rules->denyChild($this->tagName);

		// Empty this tag's filter chain and add our tag filter
		$tag->filterChain->clear();
		$tag->filterChain
		    ->append(__NAMESPACE__ . '\\Parser::filterTag')
		    ->resetParameters()
		    ->addParameterByName('tag')
		    ->addParameterByName('parser')
		    ->addParameterByName('MediaEmbed.hosts')
		    ->addParameterByName('MediaEmbed.sites')
		    ->addParameterByName('cacheDir')
		    ->setJS(file_get_contents(__DIR__ . '/Parser/tagFilter.js'));
	}

	/**
	* Return the list of named captures from a list of [regexp, map] pairs
	*
	* @param  array[] $regexps List of [regexp, map] pairs
	* @return string[]
	*/
	protected function getAttributeNamesFromRegexps(array $regexps)
	{
		$attrNames = [];
		foreach ($regexps as list($regexp, $map))
		{
			$attrNames += array_flip(array_filter($map));
		}

		return $attrNames;
	}

	/**
	* Get the attributes config for given site config
	*
	* @param  array $siteConfig Site's config
	* @return array             Map of [attrName => attrConfig]
	*/
	protected function getAttributesConfig(array $siteConfig)
	{
		$attrNames = $this->getAttributeNamesFromRegexps($siteConfig['extract']);
		foreach ($siteConfig['scrape'] as $scrapeConfig)
		{
			$attrNames += $this->getAttributeNamesFromRegexps($scrapeConfig['extract']);
		}

		$attributes = $siteConfig['attributes'] + array_fill_keys(array_keys($attrNames), []);
		foreach ($attributes as &$attrConfig)
		{
			$attrConfig += ['required' => false];
		}
		unset($attrConfig);

		return $attributes;
	}

	/**
	* Validate and normalize a site ID
	*
	* @param  string $siteId
	* @return string
	*/
	protected function normalizeId($siteId)
	{
		$siteId = strtolower($siteId);

		if (!preg_match('(^[a-z0-9]+$)', $siteId))
		{
			throw new InvalidArgumentException('Invalid site ID');
		}

		return $siteId;
	}
}

1		<?php
2
3		/**
4		* @package s9e\TextFormatter
5		* @copyright Copyright (c) The s9e authors
6		* @license http://www.opensource.org/licenses/mit-license.php The MIT License
7		*/
8		namespace s9e\TextFormatter\Plugins\MediaEmbed;
9
10		use InvalidArgumentException;
11		use RuntimeException;
12		use s9e\TextFormatter\Configurator\Helpers\FilterHelper;
13		use s9e\TextFormatter\Configurator\Items\Regexp;
14		use s9e\TextFormatter\Configurator\Items\Tag;
15		use s9e\TextFormatter\Configurator\JavaScript\Dictionary;
16		use s9e\TextFormatter\Plugins\ConfiguratorBase;
17		use s9e\TextFormatter\Plugins\MediaEmbed\Configurator\Collections\CachedDefinitionCollection;
18		use s9e\TextFormatter\Plugins\MediaEmbed\Configurator\SiteHelpers\AbstractSiteHelper;
19		use s9e\TextFormatter\Plugins\MediaEmbed\Configurator\TemplateBuilder;
20
21		class Configurator extends ConfiguratorBase
22		{
23		/**
24		* @var array List of filters that are explicitly allowed in attribute definitions
25		*/
26		public $allowedFilters = ['htmlspecialchars_decode', 'stripslashes', 'str_replace', 'urldecode'];
27
28		/**
29		* @var bool Whether to create the MEDIA BBCode
30		*/
31		protected $createMediaBBCode = true;
32
33		/**
34		* @var Configurator\Collections\SiteDefinitionCollection Default sites
35		*/
36		public $defaultSites;
37
38		/**
39		* {@inheritdoc}
40		*/
41		protected $quickMatch = '://';
42
43		/**
44		* {@inheritdoc}
45		*/
46		protected $regexp = '/\\bhttps?:\\/\\/[^["\'\\s]+/Si';
47
48		/**
49		* @var array Configured sites
50		*/
51		protected $sites = [];
52
53		/**
54		* @var string Name of the tag used to handle embeddable URLs
55		*/
56		protected $tagName = 'MEDIA';
57
58		/**
59		* @var TemplateBuilder
60		*/
61		protected $templateBuilder;
62
63		/**
64		* {@inheritdoc}
65	25	*/
66		protected function setUp()
67	25	{
68	25	$this->defaultSites = new CachedDefinitionCollection;
69		$this->templateBuilder = new TemplateBuilder;
70	25
71	25	$this->configurator->registeredVars['MediaEmbed.hosts'] = new Dictionary;
72		$this->configurator->registeredVars['MediaEmbed.sites'] = new Dictionary;
73
74	25	// Create a MEDIA tag
75		$this->createMediaTag();
76
77	25	// Create a [MEDIA] BBCode if applicable
78		if ($this->createMediaBBCode)
79	24	{
80		$this->configurator->BBCodes->set($this->tagName, ['contentAttributes' => ['url']]);
81		}
82		}
83
84		/**
85		* {@inheritdoc}
86	4	*/
87		public function asConfig()
88	4	{
89		if (empty($this->sites))
90	1	{
91		return;
92		}
93
94	3	return [
95	3	'quickMatch' => $this->quickMatch,
96	3	'regexp' => $this->regexp,
97		'tagName' => $this->tagName
98		];
99		}
100
101		/**
102		* Add a media site
103		*
104		* @param string $siteId Site's ID
105		* @param array $siteConfig Site's config
106		* @return Tag Tag created for this site
107	21	*/
108		public function add($siteId, ?array $siteConfig = null)
109		{
110	21	// Normalize or retrieve the site definition
111	20	$siteId = $this->normalizeId($siteId);
112		if (isset($siteConfig))
113	15	{
114		$siteConfig = $this->defaultSites->normalizeValue($siteConfig);
115		}
116		else
117	5	{
118		$siteConfig = $this->defaultSites->get($siteId);
119	19	}
120	19	$siteConfig['extract'] = $this->convertRegexps($siteConfig['extract']);
121		$siteConfig['scrape'] = $this->convertScrapes($siteConfig['scrape']);
122
123	19	// Check the safety of attribute filters
124		$this->checkAttributeFilters($siteConfig['attributes']);
125
126	18	// Create the tag for this site
127		$tag = $this->addTag($siteId, $siteConfig);
128
129	17	// Update the configurator's data
130	17	$this->sites[$siteId] = $siteConfig;
131		foreach ($siteConfig['host'] as $host)
132	17	{
133		$this->configurator->registeredVars['MediaEmbed.hosts'][$host] = $siteId;
134	17	}
135		$this->configurator->registeredVars['MediaEmbed.sites'][$siteId] = [$siteConfig['extract'], $siteConfig['scrape']];
136	17
137		return $tag;
138		}
139
140		public function getSiteHelper(string $siteId): AbstractSiteHelper
141		{
142		$className = $this->defaultSites->get($siteId)['helper'];
143
144	1	return new $className($this->configurator);
145		}
146	1
147		/**
148		* Return the list of configured sites
149		*
150		* @return array Site's ID as keys, site's config as values
151		*/
152		public function getSites()
153		{
154		return $this->sites;
155		}
156	18
157		/**
158	18	* Create and return a tag that handles given media site
159	18	*
160		* @param string $siteId
161	18	* @param array $siteConfig
162		* @return Tag
163	18	*/
164		protected function addTag($siteId, array $siteConfig)
165	18	{
166		$tag = new Tag([
167		'attributes' => $this->getAttributesConfig($siteConfig),
168	18	'rules' => [
169	18	'allowChild' => 'URL',
170	17	'autoClose' => true,
171		'denyChild' => [$siteId, $this->tagName]
172	17	],
173		'template' => $this->templateBuilder->build($siteId, $siteConfig)
174		]);
175
176		$this->configurator->templateNormalizer->normalizeTag($tag);
177		$this->configurator->templateChecker->checkTag($tag);
178		$this->configurator->tags->add($siteId, $tag);
179
180		return $tag;
181	19	}
182
183	19	/**
184		* Check the safety of given attributes
185	7	*
186		* @param array $attributes
187	3	* @return void
188		*/
189	6	protected function checkAttributeFilters(array $attributes)
190		{
191	6	foreach ($attributes as $attrConfig)
192		{
193	1	if (empty($attrConfig['filterChain']))
194		{
195		continue;
196		}
197		foreach ($attrConfig['filterChain'] as $filter)
198		{
199		if (!FilterHelper::isAllowed($filter, $this->allowedFilters))
200		{
201		throw new RuntimeException("Filter '$filter' is not allowed in media sites");
202		}
203		}
204		}
205	17	}
206
207	17	/**
208		* Convert given regexp to a [regexp, map] pair
209	17	*
210		* @param string $regexp Original regexp
211		* @return array [regexp, [list of captures' names]]
212		*/
213		protected function convertRegexp($regexp)
214		{
215		$regexp = new Regexp($regexp);
216
217		return [$regexp, $regexp->getCaptureNames()];
218	19	}
219
220	19	/**
221		* Convert a list of regexps
222		*
223		* @param string[] $regexps Original list
224		* @return array[] Converted list
225		*/
226		protected function convertRegexps(array $regexps)
227		{
228		return array_map($this->convertRegexp(...), $regexps);
229	1	}
230
231	1	/**
232	1	* Convert all regexps in a scraping config
233		*
234	1	* @param array $config Original config
235		* @return array Converted config
236		*/
237		protected function convertScrapeConfig(array $config)
238		{
239		$config['extract'] = $this->convertRegexps($config['extract']);
240		$config['match'] = $this->convertRegexps($config['match']);
241
242		return $config;
243	19	}
244
245	19	/**
246		* Convert all regexps in a list of scraping configs
247		*
248		* @param array[] $scrapes Original config
249		* @return array[] Converted config
250		*/
251		protected function convertScrapes(array $scrapes)
252		{
253	25	return array_map($this->convertScrapeConfig(...), $scrapes);
254		}
255	25
256		/**
257		* Create the default MEDIA tag
258	25	*
259	25	* @return void
260		*/
261		protected function createMediaTag()
262	25	{
263	25	$tag = $this->configurator->tags->add($this->tagName);
264	25
265	25	// This tag should not need to be closed and should not contain itself
266	25	$tag->rules->autoClose();
267	25	$tag->rules->denyChild($this->tagName);
268	25
269	25	// Empty this tag's filter chain and add our tag filter
270	25	$tag->filterChain->clear();
271	25	$tag->filterChain
272		->append(__NAMESPACE__ . '\\Parser::filterTag')
273		->resetParameters()
274		->addParameterByName('tag')
275		->addParameterByName('parser')
276		->addParameterByName('MediaEmbed.hosts')
277		->addParameterByName('MediaEmbed.sites')
278		->addParameterByName('cacheDir')
279		->setJS(file_get_contents(__DIR__ . '/Parser/tagFilter.js'));
280	18	}
281
282	18	/**
283	18	* Return the list of named captures from a list of [regexp, map] pairs
284		*
285	16	* @param array[] $regexps List of [regexp, map] pairs
286		* @return string[]
287		*/
288	18	protected function getAttributeNamesFromRegexps(array $regexps)
289		{
290		$attrNames = [];
291		foreach ($regexps as list($regexp, $map))
292		{
293		$attrNames += array_flip(array_filter($map));
294		}
295
296		return $attrNames;
297	18	}
298
299	18	/**
300	18	* Get the attributes config for given site config
301		*
302	1	* @param array $siteConfig Site's config
303		* @return array Map of [attrName => attrConfig]
304		*/
305	18	protected function getAttributesConfig(array $siteConfig)
306	18	{
307		$attrNames = $this->getAttributeNamesFromRegexps($siteConfig['extract']);
308	16	foreach ($siteConfig['scrape'] as $scrapeConfig)
309		{
310	18	$attrNames += $this->getAttributeNamesFromRegexps($scrapeConfig['extract']);
311		}
312	18
313		$attributes = $siteConfig['attributes'] + array_fill_keys(array_keys($attrNames), []);
314		foreach ($attributes as &$attrConfig)
315		{
316		$attrConfig += ['required' => false];
317		}
318		unset($attrConfig);
319
320		return $attributes;
321	21	}
322
323	21	/**
324		* Validate and normalize a site ID
325	21	*
326		* @param string $siteId
327	1	* @return string
328		*/
329		protected function normalizeId($siteId)
330	20	{
331		$siteId = strtolower($siteId);
332
333		if (!preg_match('(^[a-z0-9]+$)', $siteId))
334		{
335		throw new InvalidArgumentException('Invalid site ID');
336		}
337
338		return $siteId;
339		}
340		}

s9e / TextFormatter

Push — master ( a025c6...613f63 )

Configurator::getSiteHelper() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like