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

Completed

Branch — TemplateInspector (5726eb)

by Josh

created 2017-11-26 04:28 UTC

Parser::parseBBCode() D

↳ Parent: Parser

Complexity

Conditions	17
Paths	166

Size

Total Lines	80
Code Lines	32

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	306

Importance

Changes

Metric	Value
dl	0
loc	80
ccs	0
cts	32
cp	0
rs	4.7009
c	0
b	0
f	0
cc	17
eloc	32
nc	166
nop	0
crap	306

How to fix Long Method Complexity

<?php

/**
* @package   s9e\TextFormatter
* @copyright Copyright (c) 2010-2017 The s9e Authors
* @license   http://www.opensource.org/licenses/mit-license.php The MIT License
*/
namespace s9e\TextFormatter\Plugins\BBCodes;

use RuntimeException;
use s9e\TextFormatter\Parser\Tag;
use s9e\TextFormatter\Plugins\ParserBase;

class Parser extends ParserBase
{
	/**
	* @var array Attributes of the BBCode being parsed
	*/
	protected $attributes;

	/**
	* @var array Configuration for the BBCode being parsed
	*/
	protected $bbcodeConfig;

	/**
	* @var string Name of the BBCode being parsed
	*/
	protected $bbcodeName;

	/**
	* @var string Suffix of the BBCode being parsed, including its colon
	*/
	protected $bbcodeSuffix;

	/**
	* @var integer Position of the cursor in the original text
	*/
	protected $pos;

	/**
	* @var integer Position of the start of the BBCode being parsed
	*/
	protected $startPos;

	/**
	* @var string Text being parsed
	*/
	protected $text;

	/**
	* @var integer Length of the text being parsed
	*/
	protected $textLen;

	/**
	* @var string Text being parsed, normalized to uppercase
	*/
	protected $uppercaseText;

	/**
	* {@inheritdoc}
	*/
	public function parse($text, array $matches)
	{
		$this->text          = $text;
		$this->textLen       = strlen($text);
		$this->uppercaseText = '';
		foreach ($matches as $m)
		{
			$this->bbcodeName = strtoupper($m[1][0]);
			if (!isset($this->config['bbcodes'][$this->bbcodeName]))
			{
				continue;
			}
			$this->bbcodeConfig = $this->config['bbcodes'][$this->bbcodeName];
			$this->startPos     = $m[0][1];
			$this->pos          = $this->startPos + strlen($m[0][0]);

			try
			{
				$this->parseBBCode();
			}
			catch (RuntimeException $e)
			{
				// Do nothing
			}
		}
	}

	/**
	* Add the end tag that matches current BBCode
	*
	* @return Tag
	*/
	protected function addBBCodeEndTag()
	{
		return $this->parser->addEndTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos);
	}

	/**
	* Add the self-closing tag that matches current BBCode
	*
	* @return Tag
	*/
	protected function addBBCodeSelfClosingTag()
	{
		$tag = $this->parser->addSelfClosingTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos);
		$tag->setAttributes($this->attributes);

		return $tag;
	}

	/**
	* Add the start tag that matches current BBCode
	*
	* @return Tag
	*/
	protected function addBBCodeStartTag()
	{
		$tag = $this->parser->addStartTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos);
		$tag->setAttributes($this->attributes);

		return $tag;
	}

	/**
	* Parse the end tag that matches given BBCode name and suffix starting at current position
	*
	* @return Tag|null
	*/
	protected function captureEndTag()
	{
		if (empty($this->uppercaseText))
		{
			$this->uppercaseText = strtoupper($this->text);
		}
		$match     = '[/' . $this->bbcodeName . $this->bbcodeSuffix . ']';
		$endTagPos = strpos($this->uppercaseText, $match, $this->pos);
		if ($endTagPos === false)
		{
			return;
		}

		return $this->parser->addEndTag($this->getTagName(), $endTagPos, strlen($match));
	}

	/**
	* Get the tag name for current BBCode
	*
	* @return string
	*/
	protected function getTagName()
	{
		// Use the configured tagName if available, or reuse the BBCode's name otherwise
		return (isset($this->bbcodeConfig['tagName']))
		     ? $this->bbcodeConfig['tagName']
		     : $this->bbcodeName;
	}

	/**
	* Parse attributes starting at current position
	*
	* @return array Associative array of [name => value]
	*/
	protected function parseAttributes()
	{
		$firstPos = $this->pos;
		$this->attributes = [];
		while ($this->pos < $this->textLen)
		{
			$c = $this->text[$this->pos];
			if (strpos(" \n\t", $c) !== false)
			{
				++$this->pos;
				continue;
			}
			if (strpos('/]', $c) !== false)
			{
				return;
			}

			// Capture the attribute name
			$spn = strspn($this->text, 'abcdefghijklmnopqrstuvwxyz_0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ-', $this->pos);
			if ($spn)
			{
				$attrName = strtolower(substr($this->text, $this->pos, $spn));
				$this->pos += $spn;
				if ($this->pos >= $this->textLen)
				{
					// The attribute name extends to the end of the text
					throw new RuntimeException;
				}
				if ($this->text[$this->pos] !== '=')
				{
					// It's an attribute name not followed by an equal sign, ignore it
					continue;
				}
			}
			elseif ($c === '=' && $this->pos === $firstPos)
			{
				// This is the default param, e.g. [quote=foo]
				$attrName = (isset($this->bbcodeConfig['defaultAttribute']))
				          ? $this->bbcodeConfig['defaultAttribute']
				          : strtolower($this->bbcodeName);
			}
			else
			{
				throw new RuntimeException;
			}

			// Move past the = and make sure we're not at the end of the text
			if (++$this->pos >= $this->textLen)
			{
				throw new RuntimeException;
			}

			$this->attributes[$attrName] = $this->parseAttributeValue();
		}
	}

	/**
	* Parse the attribute value starting at current position
	*
	* @return string
	*/
	protected function parseAttributeValue()
	{
		// Test whether the value is in quotes
		if ($this->text[$this->pos] === '"' || $this->text[$this->pos] === "'")
		{
			return $this->parseQuotedAttributeValue();
		}

		// Capture everything up to whichever comes first:
		//  - an endline
		//  - whitespace followed by a slash and a closing bracket
		//  - a closing bracket, optionally preceded by whitespace
		//  - whitespace followed by another attribute (name followed by equal sign)
		//
		// NOTE: this is for compatibility with some forums (such as vBulletin it seems)
		//       that do not put attribute values in quotes, e.g.
		//       [quote=John Smith;123456] (quoting "John Smith" from post #123456)
		if (!preg_match('#[^\\]\\n]*?(?=\\s*(?:\\s/)?\\]|\\s+[-\\w]+=)#', $this->text, $m, null, $this->pos))
		{
			throw new RuntimeException;
		}

		$attrValue  = $m[0];
		$this->pos += strlen($attrValue);

		return $attrValue;
	}

	/**
	* Parse current BBCode
	*
	* @return void
	*/
	protected function parseBBCode()
	{
		$this->parseBBCodeSuffix();

		// Test whether this is an end tag
		if ($this->text[$this->startPos + 1] === '/')
		{
			// Test whether the tag is properly closed and whether this tag has an identifier.
			// We skip end tags that carry an identifier because they're automatically added
			// when their start tag is processed
			if (substr($this->text, $this->pos, 1) === ']' && $this->bbcodeSuffix === '')
			{
				++$this->pos;
				$this->addBBCodeEndTag();
			}

			return;
		}

		// Parse attributes and fill in the blanks with predefined attributes
		$this->parseAttributes();
		if (isset($this->bbcodeConfig['predefinedAttributes']))
		{
			$this->attributes += $this->bbcodeConfig['predefinedAttributes'];
		}

		// Test whether the tag is properly closed
		if (substr($this->text, $this->pos, 1) === ']')
		{
			++$this->pos;
		}
		else
		{
			// Test whether this is a self-closing tag
			if (substr($this->text, $this->pos, 2) === '/]')
			{
				$this->pos += 2;
				$this->addBBCodeSelfClosingTag();
			}

			return;
		}

		// Record the names of attributes that need the content of this tag
		$contentAttributes = [];
		if (isset($this->bbcodeConfig['contentAttributes']))
		{
			foreach ($this->bbcodeConfig['contentAttributes'] as $attrName)
			{
				if (!isset($this->attributes[$attrName]))
				{
					$contentAttributes[] = $attrName;
				}
			}
		}

		// Look ahead and parse the end tag that matches this tag, if applicable
		$requireEndTag = ($this->bbcodeSuffix || !empty($this->bbcodeConfig['forceLookahead']));
		$endTag = ($requireEndTag || !empty($contentAttributes)) ? $this->captureEndTag() : null;
		if (isset($endTag))
		{
			foreach ($contentAttributes as $attrName)
			{
				$this->attributes[$attrName] = substr($this->text, $this->pos, $endTag->getPos() - $this->pos);
			}
		}
		elseif ($requireEndTag)
		{
			return;
		}

		// Create this start tag
		$tag = $this->addBBCodeStartTag();

		// If an end tag was created, pair it with this start tag
		if (isset($endTag))
		{
			$tag->pairWith($endTag);
		}
	}

	/**
	* Parse the BBCode suffix starting at current position
	*
	* Used to explicitly pair specific tags together, e.g.
	*   [code:123][code]type your code here[/code][/code:123]
	*
	* @return void
	*/
	protected function parseBBCodeSuffix()
	{
		$this->bbcodeSuffix = '';
		if ($this->text[$this->pos] === ':')
		{
			// Capture the colon and the (0 or more) digits following it
			$spn = 1 + strspn($this->text, '0123456789', 1 + $this->pos);
			$this->bbcodeSuffix = substr($this->text, $this->pos, $spn);

			// Move past the suffix
			$this->pos += $spn;
		}
	}

	/**
	* Parse a quoted attribute value that starts at current offset
	*
	* @return string
	*/
	protected function parseQuotedAttributeValue()
	{
		$quote    = $this->text[$this->pos];
		$valuePos = $this->pos + 1;
		while (1)
		{
			// Look for the next quote
			$this->pos = strpos($this->text, $quote, $this->pos + 1);
			if ($this->pos === false)
			{
				// No matching quote. Apparently that string never ends...
				throw new RuntimeException;
			}

			// Test for an odd number of backslashes before this character
			$n = 0;
			do
			{
				++$n;
			}
			while ($this->text[$this->pos - $n] === '\\');

			if ($n % 2)
			{
				// If $n is odd, it means there's an even number of backslashes. We can exit this
				// loop
				break;
			}
		}

		// Unescape special characters ' " and \
		$attrValue = preg_replace(
			'#\\\\([\\\\\'"])#',
			'$1',
			substr($this->text, $valuePos, $this->pos - $valuePos)
		);

		// Skip past the closing quote
		++$this->pos;

		return $attrValue;
	}
}

1			<?php
2
3			/**
4			* @package s9e\TextFormatter
5			* @copyright Copyright (c) 2010-2017 The s9e Authors
6			* @license http://www.opensource.org/licenses/mit-license.php The MIT License
7			*/
8			namespace s9e\TextFormatter\Plugins\BBCodes;
9
10			use RuntimeException;
11			use s9e\TextFormatter\Parser\Tag;
12			use s9e\TextFormatter\Plugins\ParserBase;
13
14			class Parser extends ParserBase
15			{
16			/**
17			* @var array Attributes of the BBCode being parsed
18			*/
19			protected $attributes;
20
21			/**
22			* @var array Configuration for the BBCode being parsed
23			*/
24			protected $bbcodeConfig;
25
26			/**
27			* @var string Name of the BBCode being parsed
28			*/
29			protected $bbcodeName;
30
31			/**
32			* @var string Suffix of the BBCode being parsed, including its colon
33			*/
34			protected $bbcodeSuffix;
35
36			/**
37			* @var integer Position of the cursor in the original text
38			*/
39			protected $pos;
40
41			/**
42			* @var integer Position of the start of the BBCode being parsed
43			*/
44			protected $startPos;
45
46			/**
47			* @var string Text being parsed
48			*/
49			protected $text;
50
51			/**
52			* @var integer Length of the text being parsed
53			*/
54			protected $textLen;
55
56			/**
57			* @var string Text being parsed, normalized to uppercase
58			*/
59			protected $uppercaseText;
60
61			/**
62			* {@inheritdoc}
63			*/
64			public function parse($text, array $matches)
65			{
66			$this->text = $text;
67			$this->textLen = strlen($text);
68			$this->uppercaseText = '';
69			foreach ($matches as $m)
70			{
71			$this->bbcodeName = strtoupper($m[1][0]);
72			if (!isset($this->config['bbcodes'][$this->bbcodeName]))
73			{
74			continue;
75			}
76			$this->bbcodeConfig = $this->config['bbcodes'][$this->bbcodeName];
77			$this->startPos = $m[0][1];
78			$this->pos = $this->startPos + strlen($m[0][0]);
79
80			try
81			{
82			$this->parseBBCode();
83			}
84			catch (RuntimeException $e)
85			{
86			// Do nothing
87			}
88			}
89			}
90
91			/**
92			* Add the end tag that matches current BBCode
93			*
94			* @return Tag
95			*/
96			protected function addBBCodeEndTag()
97			{
98			return $this->parser->addEndTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos);
99			}
100
101			/**
102			* Add the self-closing tag that matches current BBCode
103			*
104			* @return Tag
105			*/
106			protected function addBBCodeSelfClosingTag()
107			{
108			$tag = $this->parser->addSelfClosingTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos);
109			$tag->setAttributes($this->attributes);
110
111			return $tag;
112			}
113
114			/**
115			* Add the start tag that matches current BBCode
116			*
117			* @return Tag
118			*/
119			protected function addBBCodeStartTag()
120			{
121			$tag = $this->parser->addStartTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos);
122			$tag->setAttributes($this->attributes);
123
124			return $tag;
125			}
126
127			/**
128			* Parse the end tag that matches given BBCode name and suffix starting at current position
129			*
130			* @return Tag\|null
131			*/
132			protected function captureEndTag()
133			{
134			if (empty($this->uppercaseText))
135			{
136			$this->uppercaseText = strtoupper($this->text);
137			}
138			$match = '[/' . $this->bbcodeName . $this->bbcodeSuffix . ']';
139			$endTagPos = strpos($this->uppercaseText, $match, $this->pos);
140			if ($endTagPos === false)
141			{
142			return;
143			}
144
145			return $this->parser->addEndTag($this->getTagName(), $endTagPos, strlen($match));
146			}
147
148			/**
149			* Get the tag name for current BBCode
150			*
151			* @return string
152			*/
153			protected function getTagName()
154			{
155			// Use the configured tagName if available, or reuse the BBCode's name otherwise
156			return (isset($this->bbcodeConfig['tagName']))
157			? $this->bbcodeConfig['tagName']
158			: $this->bbcodeName;
159			}
160
161			/**
162			* Parse attributes starting at current position
163			*
164			* @return array Associative array of [name => value]
165			*/
166			protected function parseAttributes()
167			{
168			$firstPos = $this->pos;
169			$this->attributes = [];
170			while ($this->pos < $this->textLen)
171			{
172			$c = $this->text[$this->pos];
173			if (strpos(" \n\t", $c) !== false)
174			{
175			++$this->pos;
176			continue;
177			}
178			if (strpos('/]', $c) !== false)
179			{
180			return;
181			}
182
183			// Capture the attribute name
184			$spn = strspn($this->text, 'abcdefghijklmnopqrstuvwxyz_0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ-', $this->pos);
185			if ($spn)
186			{
187			$attrName = strtolower(substr($this->text, $this->pos, $spn));
188			$this->pos += $spn;
189			if ($this->pos >= $this->textLen)
190			{
191			// The attribute name extends to the end of the text
192			throw new RuntimeException;
193			}
194			if ($this->text[$this->pos] !== '=')
195			{
196			// It's an attribute name not followed by an equal sign, ignore it
197			continue;
198			}
199			}
200			elseif ($c === '=' && $this->pos === $firstPos)
201			{
202			// This is the default param, e.g. [quote=foo]
203			$attrName = (isset($this->bbcodeConfig['defaultAttribute']))
204			? $this->bbcodeConfig['defaultAttribute']
205			: strtolower($this->bbcodeName);
206			}
207			else
208			{
209			throw new RuntimeException;
210			}
211
212			// Move past the = and make sure we're not at the end of the text
213			if (++$this->pos >= $this->textLen)
214			{
215			throw new RuntimeException;
216			}
217
218			$this->attributes[$attrName] = $this->parseAttributeValue();
219			}
220			}
221
222			/**
223			* Parse the attribute value starting at current position
224			*
225			* @return string
226			*/
227			protected function parseAttributeValue()
228			{
229			// Test whether the value is in quotes
230			if ($this->text[$this->pos] === '"' \|\| $this->text[$this->pos] === "'")
231			{
232			return $this->parseQuotedAttributeValue();
233			}
234
235			// Capture everything up to whichever comes first:
236			// - an endline
237			// - whitespace followed by a slash and a closing bracket
238			// - a closing bracket, optionally preceded by whitespace
239			// - whitespace followed by another attribute (name followed by equal sign)
240			//
241			// NOTE: this is for compatibility with some forums (such as vBulletin it seems)
242			// that do not put attribute values in quotes, e.g.
243			// [quote=John Smith;123456] (quoting "John Smith" from post #123456)
244			if (!preg_match('#[^\\]\\n]?(?=\\s(?:\\s/)?\\]\|\\s+[-\\w]+=)#', $this->text, $m, null, $this->pos))
245			{
246			throw new RuntimeException;
247			}
248
249			$attrValue = $m[0];
250			$this->pos += strlen($attrValue);
251
252			return $attrValue;
253			}
254
255			/**
256			* Parse current BBCode
257			*
258			* @return void
259			*/
260			protected function parseBBCode()
261			{
262			$this->parseBBCodeSuffix();
263
264			// Test whether this is an end tag
265			if ($this->text[$this->startPos + 1] === '/')
266			{
267			// Test whether the tag is properly closed and whether this tag has an identifier.
268			// We skip end tags that carry an identifier because they're automatically added
269			// when their start tag is processed
270			if (substr($this->text, $this->pos, 1) === ']' && $this->bbcodeSuffix === '')
271			{
272			++$this->pos;
273			$this->addBBCodeEndTag();
274			}
275
276			return;
277			}
278
279			// Parse attributes and fill in the blanks with predefined attributes
280			$this->parseAttributes();
281			if (isset($this->bbcodeConfig['predefinedAttributes']))
282			{
283			$this->attributes += $this->bbcodeConfig['predefinedAttributes'];
284			}
285
286			// Test whether the tag is properly closed
287			if (substr($this->text, $this->pos, 1) === ']')
288			{
289			++$this->pos;
290			}
291			else
292			{
293			// Test whether this is a self-closing tag
294			if (substr($this->text, $this->pos, 2) === '/]')
295			{
296			$this->pos += 2;
297			$this->addBBCodeSelfClosingTag();
298			}
299
300			return;
301			}
302
303			// Record the names of attributes that need the content of this tag
304			$contentAttributes = [];
305			if (isset($this->bbcodeConfig['contentAttributes']))
306			{
307			foreach ($this->bbcodeConfig['contentAttributes'] as $attrName)
308			{
309			if (!isset($this->attributes[$attrName]))
310			{
311			$contentAttributes[] = $attrName;
312			}
313			}
314			}
315
316			// Look ahead and parse the end tag that matches this tag, if applicable
317			$requireEndTag = ($this->bbcodeSuffix \|\| !empty($this->bbcodeConfig['forceLookahead']));
318			$endTag = ($requireEndTag \|\| !empty($contentAttributes)) ? $this->captureEndTag() : null;
319			if (isset($endTag))
320			{
321			foreach ($contentAttributes as $attrName)
322			{
323			$this->attributes[$attrName] = substr($this->text, $this->pos, $endTag->getPos() - $this->pos);
324			}
325			}
326			elseif ($requireEndTag)
327			{
328			return;
329			}
330
331			// Create this start tag
332			$tag = $this->addBBCodeStartTag();
333
334			// If an end tag was created, pair it with this start tag
335			if (isset($endTag))
336			{
337			$tag->pairWith($endTag);
338			}
339			}
340
341			/**
342			* Parse the BBCode suffix starting at current position
343			*
344			* Used to explicitly pair specific tags together, e.g.
345			* [code:123][code]type your code here[/code][/code:123]
346			*
347			* @return void
348			*/
349			protected function parseBBCodeSuffix()
350			{
351			$this->bbcodeSuffix = '';
352			if ($this->text[$this->pos] === ':')
353			{
354			// Capture the colon and the (0 or more) digits following it
355			$spn = 1 + strspn($this->text, '0123456789', 1 + $this->pos);
356			$this->bbcodeSuffix = substr($this->text, $this->pos, $spn);
357
358			// Move past the suffix
359			$this->pos += $spn;
360			}
361			}
362
363			/**
364			* Parse a quoted attribute value that starts at current offset
365			*
366			* @return string
367			*/
368			protected function parseQuotedAttributeValue()
369			{
370			$quote = $this->text[$this->pos];
371			$valuePos = $this->pos + 1;
372			while (1)
373			{
374			// Look for the next quote
375			$this->pos = strpos($this->text, $quote, $this->pos + 1);
376			if ($this->pos === false)
377			{
378			// No matching quote. Apparently that string never ends...
379			throw new RuntimeException;
380			}
381
382			// Test for an odd number of backslashes before this character
383			$n = 0;
384			do
385			{
386			++$n;
387			}
388			while ($this->text[$this->pos - $n] === '\\');
389
390			if ($n % 2)
391			{
392			// If $n is odd, it means there's an even number of backslashes. We can exit this
393			// loop
394			break;
395			}
396			}
397
398			// Unescape special characters ' " and \
399			$attrValue = preg_replace(
400			'#\\\\([\\\\\'"])#',
401			'$1',
402			substr($this->text, $valuePos, $this->pos - $valuePos)
403			);
404
405			// Skip past the closing quote
406			++$this->pos;
407
408			return $attrValue;
409			}
410			}

s9e / TextFormatter

Branch — TemplateInspector (5726eb)

Parser::parseBBCode() D

Complexity

Size

Duplication

Code Coverage

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like