XPathHelper::getVariables() - Code Metrics - Inspection of "XPathHelper: ignore malformed expression in minify..." - s9e/TextFormatter - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( aec2d9...ec72f4 )

by Josh

created 2020-01-22 03:26 UTC

XPathHelper::getVariables() A

↳ Parent: XPathHelper

Complexity

Conditions	1
Paths	1

Size

Total Lines	13
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	6
CRAP Score	1

Importance

Changes

Metric	Value
eloc	5
dl	0
loc	13
ccs	6
cts	6
cp	1
rs	10
c	0
b	0
f	0
cc	1
nc	1
nop	1
crap	1

<?php

/**
* @package   s9e\TextFormatter
* @copyright Copyright (c) 2010-2020 The s9e authors
* @license   http://www.opensource.org/licenses/mit-license.php The MIT License
*/
namespace s9e\TextFormatter\Configurator\Helpers;

use RuntimeException;
use s9e\TextFormatter\Configurator\RecursiveParser;
use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\BooleanFunctions;
use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\BooleanOperators;
use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\Comparisons;
use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\Core;
use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\Math;
use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\SingleByteStringFunctions;
use s9e\TextFormatter\Utils\XPath;

abstract class XPathHelper
{
	/**
	* Decode strings inside of an XPath expression
	*
	* @param  string $expr
	* @return string
	*/
	public static function decodeStrings($expr)
	{
		return preg_replace_callback(
			'(\'[^\']*+\'|"[^"]*+")',
			function ($m)
			{
				return $m[0][0] . hex2bin(substr($m[0], 1, -1)) . $m[0][0];
			},
			$expr
		);
	}

	/**
	* Encode strings inside of an XPath expression
	*
	* @param  string $expr
	* @return string
	*/
	public static function encodeStrings($expr)
	{
		return preg_replace_callback(
			'(\'[^\']*+\'|"[^"]*+")',
			function ($m)
			{
				return $m[0][0] . bin2hex(substr($m[0], 1, -1)) . $m[0][0];
			},
			$expr
		);
	}

	/**
	* Return the list of variables used in a given XPath expression
	*
	* @param  string $expr XPath expression
	* @return array        Alphabetically sorted list of unique variable names
	*/
	public static function getVariables($expr)
	{
		// First, remove strings' contents to prevent false-positives
		$expr = preg_replace('/(["\']).*?\\1/s', '$1$1', $expr);

		// Capture all the variable names
		preg_match_all('/\\$(\\w+)/', $expr, $matches);

		// Dedupe and sort names
		$varNames = array_unique($matches[1]);
		sort($varNames);

		return $varNames;
	}

	/**
	* Determine whether given XPath expression definitely evaluates to a number
	*
	* @param  string $expr XPath expression
	* @return bool         Whether given XPath expression definitely evaluates to a number
	*/
	public static function isExpressionNumeric($expr)
	{
		// Detect simple arithmetic operations
		if (preg_match('(^([$@][-\\w]++|-?[.\\d]++)(?: *(?:[-*+]|div) *(?1))+$)', $expr))
		{
			return true;
		}

		// Try parsing the expression as a math expression
		try
		{
			return (bool) self::getXPathParser()->parse($expr, 'Math');
		}
		catch (RuntimeException $e)
		{
			// Do nothing
		}

		return false;
	}

	/**
	* Remove extraneous space in a given XPath expression
	*
	* @param  string $expr Original XPath expression
	* @return string       Minified XPath expression
	*/
	public static function minify($expr)
	{
		$expr = trim($expr);

		// Test whether there's any characters that can be removed
		if (!preg_match('([\\s\\)])', $expr))
		{
			return $expr;
		}

		// Temporarily encode the content of literal strings
		$expr = self::encodeStrings(trim($expr));

		// Normalize whitespace to a single space
		$expr = preg_replace('(\\s+)', ' ', $expr);

		$regexps = [
			// Remove the space between a non-word character and a word character
			'([-a-z_0-9]\\K (?=[^-a-z_0-9]))i',
			'([^-a-z_0-9]\\K (?=[-a-z_0-9]))i',

			// Remove the space between two non-word characters as long as they're not two -
			'((?!- -)[^-a-z_0-9]\\K (?=[^-a-z_0-9]))i',

			// Remove the space between a - and a word character as long as there's a space before -
			'( -\\K (?=[a-z_0-9]))i',

			// Remove the space between an operator and the next token
			'([ \\)](?:and|div|or)\\K )',

			// Remove the space after a number
			'(\\b\\d+\\K )'
		];
		$expr = preg_replace($regexps, '', $expr);

		// Remove consecutive parentheses where redundant
		$expr = self::removeRedundantParentheses($expr);

		// Restore the literals
		$expr = self::decodeStrings($expr);

		return $expr;
	}

	/**
	* Remove consecutive parentheses where redundant
	*/
	protected static function removeRedundantParentheses(string $expr): string
	{
		// Add parentheses around the original expression and terminate the expression with a space
		preg_match_all('([\\(\\)]|[^\\(\\)]++)', '(' . $expr . ') ', $m);
		$tokens = $m[0];

		$depth = 0;
		$left  = [-1 => null];
		foreach ($tokens as $k => $token)
		{
			if ($token === '(')
			{
				$left[$depth++] = $k;
			}
			elseif ($token === ')' && --$depth >= 0 && $tokens[$k + 1] === ')' && $left[$depth - 1] === $left[$depth] - 1)
			{
				unset($tokens[$k], $tokens[$left[$depth]]);
			}
		}

		// Remove the extra parentheses as well as the last token before serializing them
		return implode('', array_slice($tokens, 1, -2));
	}

	/**
	* Parse an XPath expression that is composed entirely of equality tests between a variable part
	* and a constant part
	*
	* @param  string      $expr
	* @return array|false
	*/
	public static function parseEqualityExpr($expr)
	{
		// Match an equality between a variable and a literal or the concatenation of strings
		$eq = '(?<equality>'
		    . '(?<key>@[-\\w]+|\\$\\w+|\\.)'
		    . '(?<operator>\\s*=\\s*)'
		    . '(?:'
		    . '(?<literal>(?<string>"[^"]*"|\'[^\']*\')|0|[1-9][0-9]*)'
		    . '|'
		    . '(?<concat>concat\\(\\s*(?&string)\\s*(?:,\\s*(?&string)\\s*)+\\))'
		    . ')'
		    . '|'
		    . '(?:(?<literal>(?&literal))|(?<concat>(?&concat)))(?&operator)(?<key>(?&key))'
		    . ')';

		// Match a string that is entirely composed of equality checks separated with "or"
		$regexp = '(^(?J)\\s*' . $eq . '\\s*(?:or\\s*(?&equality)\\s*)*$)';
		if (!preg_match($regexp, $expr))
		{
			return false;
		}

		preg_match_all("((?J)$eq)", $expr, $matches, PREG_SET_ORDER);

		$map = [];
		foreach ($matches as $m)
		{
			$key   = $m['key'];
			$value = (!empty($m['concat']))
			       ? self::evaluateConcat($m['concat'])
			       : self::evaluateLiteral($m['literal']);

			$map[$key][] = $value;
		}

		return $map;
	}

	/**
	* Evaluate a concat() expression where all arguments are string literals
	*
	* @param  string $expr concat() expression
	* @return string       Expression's value
	*/
	protected static function evaluateConcat($expr)
	{
		preg_match_all('(\'[^\']*\'|"[^"]*")', $expr, $strings);

		$value = '';
		foreach ($strings[0] as $string)
		{
			$value .= substr($string, 1, -1);
		}

		return $value;
	}

	/**
	* Evaluate an XPath literal
	*
	* @param  string $expr XPath literal
	* @return string       Literal's string value
	*/
	protected static function evaluateLiteral($expr)
	{
		if ($expr[0] === '"' || $expr[0] === "'")
		{
			$expr = substr($expr, 1, -1);
		}

		return $expr;
	}

	/**
	* Generate and return a cached XPath parser with a default set of matchers
	*
	* @return RecursiveParser
	*/
	protected static function getXPathParser()
	{
		static $parser;
		if (!isset($parser))
		{
			$parser     = new RecursiveParser;
			$matchers   = [];
			$matchers[] = new BooleanFunctions($parser);
			$matchers[] = new BooleanOperators($parser);
			$matchers[] = new Comparisons($parser);
			$matchers[] = new Core($parser);
			$matchers[] = new Math($parser);
			$matchers[] = new SingleByteStringFunctions($parser);

			$parser->setMatchers($matchers);
		}

		return $parser;
	}
}

1		<?php
2
3		/**
4		* @package s9e\TextFormatter
5		* @copyright Copyright (c) 2010-2020 The s9e authors
6		* @license http://www.opensource.org/licenses/mit-license.php The MIT License
7		*/
8		namespace s9e\TextFormatter\Configurator\Helpers;
9
10		use RuntimeException;
11		use s9e\TextFormatter\Configurator\RecursiveParser;
12		use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\BooleanFunctions;
13		use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\BooleanOperators;
14		use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\Comparisons;
15		use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\Core;
16		use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\Math;
17		use s9e\TextFormatter\Configurator\RendererGenerators\PHP\XPathConvertor\Convertors\SingleByteStringFunctions;
18		use s9e\TextFormatter\Utils\XPath;
19
20		abstract class XPathHelper
21		{
22		/**
23		* Decode strings inside of an XPath expression
24		*
25		* @param string $expr
26		* @return string
27		*/
28	27	public static function decodeStrings($expr)
29		{
30	27	return preg_replace_callback(
31	27	'(\'[^\']+\'\|"[^"]+")',
32		function ($m)
33		{
34	3	return $m[0][0] . hex2bin(substr($m[0], 1, -1)) . $m[0][0];
35	27	},
36		$expr
37		);
38		}
39
40		/**
41		* Encode strings inside of an XPath expression
42		*
43		* @param string $expr
44		* @return string
45		*/
46	27	public static function encodeStrings($expr)
47		{
48	27	return preg_replace_callback(
49	27	'(\'[^\']+\'\|"[^"]+")',
50		function ($m)
51		{
52	3	return $m[0][0] . bin2hex(substr($m[0], 1, -1)) . $m[0][0];
53	27	},
54		$expr
55		);
56		}
57
58		/**
59		* Return the list of variables used in a given XPath expression
60		*
61		* @param string $expr XPath expression
62		* @return array Alphabetically sorted list of unique variable names
63		*/
64	4	public static function getVariables($expr)
65		{
66		// First, remove strings' contents to prevent false-positives
67	4	$expr = preg_replace('/(["\']).*?\\1/s', '$1$1', $expr);
68
69		// Capture all the variable names
70	4	preg_match_all('/\\$(\\w+)/', $expr, $matches);
71
72		// Dedupe and sort names
73	4	$varNames = array_unique($matches[1]);
74	4	sort($varNames);
75
76	4	return $varNames;
77		}
78
79		/**
80		* Determine whether given XPath expression definitely evaluates to a number
81		*
82		* @param string $expr XPath expression
83		* @return bool Whether given XPath expression definitely evaluates to a number
84		*/
85	21	public static function isExpressionNumeric($expr)
86		{
87		// Detect simple arithmetic operations
88	21	if (preg_match('(^([$@][-\\w]++\|-?[.\\d]++)(?: (?:[-+]\|div) *(?1))+$)', $expr))
89		{
90	12	return true;
91		}
92
93		// Try parsing the expression as a math expression
94		try
95		{
96	9	return (bool) self::getXPathParser()->parse($expr, 'Math');
97		}
98	5	catch (RuntimeException $e)
99		{
100		// Do nothing
101		}
102
103	5	return false;
104		}
105
106		/**
107		* Remove extraneous space in a given XPath expression
108		*
109		* @param string $expr Original XPath expression
110		* @return string Minified XPath expression
111		*/
112	28	public static function minify($expr)
113		{
114	28	$expr = trim($expr);
115
116		// Test whether there's any characters that can be removed
117	28	if (!preg_match('([\\s\\)])', $expr))
118		{
119	2	return $expr;
120		}
121
122		// Temporarily encode the content of literal strings
123	26	$expr = self::encodeStrings(trim($expr));
124
125		// Normalize whitespace to a single space
126	26	$expr = preg_replace('(\\s+)', ' ', $expr);
127
128		$regexps = [
129		// Remove the space between a non-word character and a word character
130	26	'([-a-z_0-9]\\K (?=[^-a-z_0-9]))i',
131		'([^-a-z_0-9]\\K (?=[-a-z_0-9]))i',
132
133		// Remove the space between two non-word characters as long as they're not two -
134		'((?!- -)[^-a-z_0-9]\\K (?=[^-a-z_0-9]))i',
135
136		// Remove the space between a - and a word character as long as there's a space before -
137		'( -\\K (?=[a-z_0-9]))i',
138
139		// Remove the space between an operator and the next token
140		'([ \\)](?:and\|div\|or)\\K )',
141
142		// Remove the space after a number
143		'(\\b\\d+\\K )'
144		];
145	26	$expr = preg_replace($regexps, '', $expr);
146
147		// Remove consecutive parentheses where redundant
148	26	$expr = self::removeRedundantParentheses($expr);
149
150		// Restore the literals
151	26	$expr = self::decodeStrings($expr);
152
153	26	return $expr;
154		}
155
156		/**
157		* Remove consecutive parentheses where redundant
158		*/
159	26	protected static function removeRedundantParentheses(string $expr): string
160		{
161		// Add parentheses around the original expression and terminate the expression with a space
162	26	preg_match_all('([\\(\\)]\|[^\\(\\)]++)', '(' . $expr . ') ', $m);
163	26	$tokens = $m[0];
164
165	26	$depth = 0;
166	26	$left = [-1 => null];
167	26	foreach ($tokens as $k => $token)
168		{
169	26	if ($token === '(')
170		{
171	26	$left[$depth++] = $k;
172		}
173	26	elseif ($token === ')' && --$depth >= 0 && $tokens[$k + 1] === ')' && $left[$depth - 1] === $left[$depth] - 1)
174		{
175	4	unset($tokens[$k], $tokens[$left[$depth]]);
176		}
177		}
178
179		// Remove the extra parentheses as well as the last token before serializing them
180	26	return implode('', array_slice($tokens, 1, -2));
181		}
182
183		/**
184		* Parse an XPath expression that is composed entirely of equality tests between a variable part
185		* and a constant part
186		*
187		* @param string $expr
188		* @return array\|false
189		*/
190	8	public static function parseEqualityExpr($expr)
191		{
192		// Match an equality between a variable and a literal or the concatenation of strings
193		$eq = '(?<equality>'
194		. '(?<key>@[-\\w]+\|\\$\\w+\|\\.)'
195		. '(?<operator>\\s=\\s)'
196		. '(?:'
197		. '(?<literal>(?<string>"[^"]"\|\'[^\']\')\|0\|[1-9][0-9]*)'
198		. '\|'
199		. '(?<concat>concat\\(\\s(?&string)\\s(?:,\\s(?&string)\\s)+\\))'
200		. ')'
201		. '\|'
202		. '(?:(?<literal>(?&literal))\|(?<concat>(?&concat)))(?&operator)(?<key>(?&key))'
203	8	. ')';
204
205		// Match a string that is entirely composed of equality checks separated with "or"
206	8	$regexp = '(^(?J)\\s' . $eq . '\\s(?:or\\s(?&equality)\\s)*$)';
207	8	if (!preg_match($regexp, $expr))
208		{
209	1	return false;
210		}
211
212	7	preg_match_all("((?J)$eq)", $expr, $matches, PREG_SET_ORDER);
213
214	7	$map = [];
215	7	foreach ($matches as $m)
216		{
217	7	$key = $m['key'];
218	7	$value = (!empty($m['concat']))
219	1	? self::evaluateConcat($m['concat'])
220	7	: self::evaluateLiteral($m['literal']);
221
222	7	$map[$key][] = $value;
223		}
224
225	7	return $map;
226		}
227
228		/**
229		* Evaluate a concat() expression where all arguments are string literals
230		*
231		* @param string $expr concat() expression
232		* @return string Expression's value
233		*/
234	1	protected static function evaluateConcat($expr)
235		{
236	1	preg_match_all('(\'[^\']\'\|"[^"]")', $expr, $strings);
237
238	1	$value = '';
239	1	foreach ($strings[0] as $string)
240		{
241	1	$value .= substr($string, 1, -1);
242		}
243
244	1	return $value;
245		}
246
247		/**
248		* Evaluate an XPath literal
249		*
250		* @param string $expr XPath literal
251		* @return string Literal's string value
252		*/
253	6	protected static function evaluateLiteral($expr)
254		{
255	6	if ($expr[0] === '"' \|\| $expr[0] === "'")
256		{
257	6	$expr = substr($expr, 1, -1);
258		}
259
260	6	return $expr;
261		}
262
263		/**
264		* Generate and return a cached XPath parser with a default set of matchers
265		*
266		* @return RecursiveParser
267		*/
268	9	protected static function getXPathParser()
269		{
270	9	static $parser;
271	9	if (!isset($parser))
272		{
273	1	$parser = new RecursiveParser;
274	1	$matchers = [];
275	1	$matchers[] = new BooleanFunctions($parser);
276	1	$matchers[] = new BooleanOperators($parser);
277	1	$matchers[] = new Comparisons($parser);
278	1	$matchers[] = new Core($parser);
279	1	$matchers[] = new Math($parser);
280	1	$matchers[] = new SingleByteStringFunctions($parser);
281
282	1	$parser->setMatchers($matchers);
283		}
284
285	9	return $parser;
286		}
287		}

s9e / TextFormatter

Push — master ( aec2d9...ec72f4 )

XPathHelper::getVariables() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like