Parser::parseQuotedAttributeValue() - Code Metrics - s9e/TextFormatter - Measure and Improve Code Quality continuously with Scrutinizer

Parser::parseQuotedAttributeValue() A
last analyzed 2025-10-26 07:39 UTC

↳ Parent: Parser

Complexity

Conditions	5
Paths	5

Size

Total Lines	33
Code Lines	15

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	14
CRAP Score	5

Importance

Changes

Metric	Value
eloc	15
dl	0
loc	33
ccs	14
cts	14
cp	1
rs	9.4555
c	0
b	0
f	0
cc	5
nc	5
nop	0
crap	5

<?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\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()
	{
		$prio = ($this->bbcodeSuffix !== '') ? -10 : 0;
		$tag  = $this->parser->addStartTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos, $prio);
		$tag->setAttributes($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 void
	*/
	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)
		preg_match('((?:[^\\s\\]]|[ \\t](?!\\s*(?:[-\\w]+=|/?\\])))*)', $this->text, $m, 0, $this->pos);

		$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
		$this->parseAttributes();

		// 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;
		do
		{
			// 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
			$n = 1;
			while ($this->text[$this->pos - $n] === '\\')
			{
				++$n;
			}
		}
		while ($n % 2 === 0);

		$attrValue = substr($this->text, $valuePos, $this->pos - $valuePos);
		if (strpos($attrValue, '\\') !== false)
		{
			$attrValue = strtr($attrValue, ['\\\\' => '\\', '\\"' => '"', "\\'" => "'"]);
		}

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

		return $attrValue;
	}
}

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\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 $uppercaseText;
60
61		/**
62		* {@inheritdoc}
63		*/
64	52	public function parse($text, array $matches)
65		{
66	52	$this->text = $text;
67	52	$this->textLen = strlen($text);
68	52	$this->uppercaseText = '';
69	52	foreach ($matches as $m)
70		{
71	52	$this->bbcodeName = strtoupper($m[1][0]);
72	52	if (!isset($this->config['bbcodes'][$this->bbcodeName]))
73		{
74	3	continue;
75		}
76	52	$this->bbcodeConfig = $this->config['bbcodes'][$this->bbcodeName];
77	52	$this->startPos = $m[0][1];
78	52	$this->pos = $this->startPos + strlen($m[0][0]);
79
80		try
81		{
82	52	$this->parseBBCode();
83		}
84	6	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	24	protected function addBBCodeEndTag()
97		{
98	24	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	13	protected function addBBCodeSelfClosingTag()
107		{
108	13	$tag = $this->parser->addSelfClosingTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos);
109	13	$tag->setAttributes($this->attributes);
110
111	13	return $tag;
112		}
113
114		/**
115		* Add the start tag that matches current BBCode
116		*
117		* @return Tag
118		*/
119	25	protected function addBBCodeStartTag()
120		{
121	25	$prio = ($this->bbcodeSuffix !== '') ? -10 : 0;
122	25	$tag = $this->parser->addStartTag($this->getTagName(), $this->startPos, $this->pos - $this->startPos, $prio);
123		$tag->setAttributes($this->attributes);
124	25
125		return $tag;
126		}
127
128		/**
129		* Parse the end tag that matches given BBCode name and suffix starting at current position
130		*
131		* @return Tag\|null
132	12	*/
133		protected function captureEndTag()
134	12	{
135		if (empty($this->uppercaseText))
136	12	{
137		$this->uppercaseText = strtoupper($this->text);
138	12	}
139	12	$match = '[/' . $this->bbcodeName . $this->bbcodeSuffix . ']';
140	12	$endTagPos = strpos($this->uppercaseText, $match, $this->pos);
141		if ($endTagPos === false)
142	5	{
143		return;
144		}
145	8
146		return $this->parser->addEndTag($this->getTagName(), $endTagPos, strlen($match));
147		}
148
149		/**
150		* Get the tag name for current BBCode
151		*
152		* @return string
153	39	*/
154		protected function getTagName()
155		{
156	39	// Use the configured tagName if available, or reuse the BBCode's name otherwise
157	1	return (isset($this->bbcodeConfig['tagName']))
158	39	? $this->bbcodeConfig['tagName']
159		: $this->bbcodeName;
160		}
161
162		/**
163		* Parse attributes starting at current position
164		*
165		* @return void
166	51	*/
167		protected function parseAttributes()
168	51	{
169	51	$firstPos = $this->pos;
170	51	$this->attributes = [];
171		while ($this->pos < $this->textLen)
172	51	{
173	51	$c = $this->text[$this->pos];
174		if (strpos(" \n\t", $c) !== false)
175	26	{
176	26	++$this->pos;
177		continue;
178	50	}
179		if (strpos('/]', $c) !== false)
180	43	{
181		return;
182		}
183
184	29	// Capture the attribute name
185	29	$spn = strspn($this->text, 'abcdefghijklmnopqrstuvwxyz_0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ-', $this->pos);
186		if ($spn)
187	20	{
188	20	$attrName = strtolower(substr($this->text, $this->pos, $spn));
189	20	$this->pos += $spn;
190		if ($this->pos >= $this->textLen)
191		{
192	1	// The attribute name extends to the end of the text
193		throw new RuntimeException;
194	19	}
195		if ($this->text[$this->pos] !== '=')
196		{
197	19	// It's an attribute name not followed by an equal sign, ignore it
198		continue;
199		}
200	10	}
201		elseif ($c === '=' && $this->pos === $firstPos)
202		{
203	9	// This is the default param, e.g. [quote=foo]
204	2	$attrName = (isset($this->bbcodeConfig['defaultAttribute']))
205	9	? $this->bbcodeConfig['defaultAttribute']
206		: strtolower($this->bbcodeName);
207		}
208		else
209	1	{
210		throw new RuntimeException;
211		}
212
213	26	// Move past the = and make sure we're not at the end of the text
214		if (++$this->pos >= $this->textLen)
215	1	{
216		throw new RuntimeException;
217		}
218	25
219		$this->attributes[$attrName] = $this->parseAttributeValue();
220		}
221		}
222
223		/**
224		* Parse the attribute value starting at current position
225		*
226		* @return string
227	25	*/
228		protected function parseAttributeValue()
229		{
230	25	// Test whether the value is in quotes
231		if ($this->text[$this->pos] === '"' \|\| $this->text[$this->pos] === "'")
232	14	{
233		return $this->parseQuotedAttributeValue();
234		}
235
236		// Capture everything up to whichever comes first:
237		// - an endline
238		// - whitespace followed by a slash and a closing bracket
239		// - a closing bracket, optionally preceded by whitespace
240		// - whitespace followed by another attribute (name followed by equal sign)
241		//
242		// NOTE: this is for compatibility with some forums (such as vBulletin it seems)
243		// that do not put attribute values in quotes, e.g.
244	11	// [quote=John Smith;123456] (quoting "John Smith" from post #123456)
245		preg_match('((?:[^\\s\\]]\|[ \\t](?!\\s(?:[-\\w]+=\|/?\\]))))', $this->text, $m, 0, $this->pos);
246	11
247	11	$attrValue = $m[0];
248		$this->pos += strlen($attrValue);
249	11
250		return $attrValue;
251		}
252
253		/**
254		* Parse current BBCode
255		*
256		* @return void
257	52	*/
258		protected function parseBBCode()
259	52	{
260		$this->parseBBCodeSuffix();
261
262	52	// Test whether this is an end tag
263		if ($this->text[$this->startPos + 1] === '/')
264		{
265		// Test whether the tag is properly closed and whether this tag has an identifier.
266		// We skip end tags that carry an identifier because they're automatically added
267	27	// when their start tag is processed
268		if (substr($this->text, $this->pos, 1) === ']' && $this->bbcodeSuffix === '')
269	24	{
270	24	++$this->pos;
271		$this->addBBCodeEndTag();
272		}
273	27
274		return;
275		}
276
277	51	// Parse attributes
278		$this->parseAttributes();
279
280	45	// Test whether the tag is properly closed
281		if (substr($this->text, $this->pos, 1) === ']')
282	28	{
283		++$this->pos;
284		}
285		else
286		{
287	17	// Test whether this is a self-closing tag
288		if (substr($this->text, $this->pos, 2) === '/]')
289	13	{
290	13	$this->pos += 2;
291		$this->addBBCodeSelfClosingTag();
292		}
293	17
294		return;
295		}
296
297	28	// Record the names of attributes that need the content of this tag
298	28	$contentAttributes = [];
299		if (isset($this->bbcodeConfig['contentAttributes']))
300	5	{
301		foreach ($this->bbcodeConfig['contentAttributes'] as $attrName)
302	5	{
303		if (!isset($this->attributes[$attrName]))
304	4	{
305		$contentAttributes[] = $attrName;
306		}
307		}
308		}
309
310	28	// Look ahead and parse the end tag that matches this tag, if applicable
311	28	$requireEndTag = ($this->bbcodeSuffix \|\| !empty($this->bbcodeConfig['forceLookahead']));
312	28	$endTag = ($requireEndTag \|\| !empty($contentAttributes)) ? $this->captureEndTag() : null;
313		if (isset($endTag))
314	8	{
315		foreach ($contentAttributes as $attrName)
316	3	{
317		$this->attributes[$attrName] = substr($this->text, $this->pos, $endTag->getPos() - $this->pos);
318		}
319	23	}
320		elseif ($requireEndTag)
321	4	{
322		return;
323		}
324
325	25	// Create this start tag
326		$tag = $this->addBBCodeStartTag();
327
328	25	// If an end tag was created, pair it with this start tag
329		if (isset($endTag))
330	8	{
331		$tag->pairWith($endTag);
332		}
333		}
334
335		/**
336		* Parse the BBCode suffix starting at current position
337		*
338		* Used to explicitly pair specific tags together, e.g.
339		* [code:123][code]type your code here[/code][/code:123]
340		*
341		* @return void
342	52	*/
343		protected function parseBBCodeSuffix()
344	52	{
345	52	$this->bbcodeSuffix = '';
346		if ($this->text[$this->pos] === ':')
347		{
348	7	// Capture the colon and the (0 or more) digits following it
349	7	$spn = 1 + strspn($this->text, '0123456789', 1 + $this->pos);
350		$this->bbcodeSuffix = substr($this->text, $this->pos, $spn);
351
352	7	// Move past the suffix
353		$this->pos += $spn;
354		}
355		}
356
357		/**
358		* Parse a quoted attribute value that starts at current offset
359		*
360		* @return string
361	14	*/
362		protected function parseQuotedAttributeValue()
363	14	{
364	14	$quote = $this->text[$this->pos];
365		$valuePos = $this->pos + 1;
366		do
367		{
368	14	// Look for the next quote
369	14	$this->pos = strpos($this->text, $quote, $this->pos + 1);
370		if ($this->pos === false)
371		{
372	3	// No matching quote. Apparently that string never ends...
373		throw new RuntimeException;
374		}
375
376	11	// Test for an odd number of backslashes before this character
377	11	$n = 1;
378		while ($this->text[$this->pos - $n] === '\\')
379	3	{
380		++$n;
381		}
382	11	}
383		while ($n % 2 === 0);
384	11
385	11	$attrValue = substr($this->text, $valuePos, $this->pos - $valuePos);
386		if (strpos($attrValue, '\\') !== false)
387	3	{
388		$attrValue = strtr($attrValue, ['\\\\' => '\\', '\\"' => '"', "\\'" => "'"]);
389		}
390
391	11	// Skip past the closing quote
392		++$this->pos;
393	11
394		return $attrValue;
395		}
396		}

s9e / TextFormatter

Parser::parseQuotedAttributeValue() A last analyzed 2025-10-26 07:39 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

Parser::parseQuotedAttributeValue() A
last analyzed 2025-10-26 07:39 UTC