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

Completed

Branch — 2.3/RemoveRedundantParentheses... (567a0c)

by Josh

created 2020-01-11 02:19 UTC

XPathHelper::parseEqualityExpr() A

↳ Parent: XPathHelper

Complexity

Conditions	4
Paths	4

Size

Total Lines	36
Code Lines	23

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	14
CRAP Score	4

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
eloc	23
dl	0
loc	36
ccs	14
cts	14
cp	1
rs	9.552
c	1
b	0
f	0
cc	4
nc	4
nop	1
crap	4

<?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)
	{
		preg_match_all('("[^"]*+"|\'[^\']*+\'|[\'"](*:X))', $expr, $m);
		if (!empty($m['MARK']))
		{
			throw new RuntimeException("Cannot parse XPath expression '" . $expr . "'");
		}

		// Trim the surrounding whitespace then temporarily remove literal strings
		$expr = self::encodeStrings(trim($expr));

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

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

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

		// Remove the space between a - and a word character, as long as there's a space before -
		$expr = preg_replace('/ - ([a-z_0-9])/i', ' -$1', $expr);

		// Remove the spaces between a number and a div or "-" operator and the next token
		$expr = preg_replace('/(?:^|[ \\(])\\d+\\K (div|-) ?/', '$1', $expr);

		// Remove the space between the div operator the next token
		$expr = preg_replace('/([^-a-z_0-9]div) (?=[$0-9@])/', '$1', $expr);

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

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

		return $expr;
	}

	/**
	* Remove consecutive parentheses where redundant
	*
	* @param  string $expr
	* @return string
	*/
	protected static function removeRedundantParentheses(string $expr): string
	{
		preg_match_all('([\\(\\)]|[^\\(\\)]++)', $expr, $m);
		$tokens   = $m[0];
		$tokens[] = '';

		$depth = 0;
		$left  = [-1 => -1];
		foreach ($tokens as $k => $token)
		{
			if ($token === '(')
			{
				$left[$depth] = $k;
				++$depth;
			}
			elseif ($token === ')')
			{
				if (--$depth < 0)
				{
					throw new RuntimeException("Cannot parse XPath expression '" . $expr . "'");
				}
				if ($tokens[$k + 1] === ')' && $left[$depth - 1] === $left[$depth] - 1)
				{
					unset($tokens[$k]);
					unset($tokens[$left[$depth]]);
				}
			}
		}

		return implode('', $tokens);
	}

	/**
	* 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 $expr
26		* @return string
27		*/
28	21	public static function decodeStrings($expr)
29		{
30	21	return preg_replace_callback(
31	21	'(\'[^\']+\'\|"[^"]+")',
32		function ($m)
33		{
34	2	return $m[0][0] . hex2bin(substr($m[0], 1, -1)) . $m[0][0];
35	21	},
36		$expr
37		);
38		}
39
40		/**
41		* Encode strings inside of an XPath expression
42		*
43		* @param string $expr
44		* @return string
45		*/
46	21	public static function encodeStrings($expr)
47		{
48	21	return preg_replace_callback(
49	21	'(\'[^\']+\'\|"[^"]+")',
50		function ($m)
51		{
52	2	return $m[0][0] . bin2hex(substr($m[0], 1, -1)) . $m[0][0];
53	21	},
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 $expr Original XPath expression
110		* @return string Minified XPath expression
111		*/
112	21	public static function minify($expr)
113		{
114	21	preg_match_all('("[^"]+"\|\'[^\']+\'\|[\'"](*:X))', $expr, $m);
115	21	if (!empty($m['MARK']))
116		{
117	1	throw new RuntimeException("Cannot parse XPath expression '" . $expr . "'");
118		}
119
120		// Trim the surrounding whitespace then temporarily remove literal strings
121	20	$expr = self::encodeStrings(trim($expr));
122
123		// Normalize whitespace to a single space
124	20	$expr = preg_replace('/\\s+/', ' ', $expr);
125
126		// Remove the space between a non-word character and a word character
127	20	$expr = preg_replace('/[-a-z_0-9]\\K (?=[^-a-z_0-9])/i', '', $expr);
128	20	$expr = preg_replace('/[^-a-z_0-9]\\K (?=[-a-z_0-9])/i', '', $expr);
129
130		// Remove the space between two non-word characters as long as they're not two -
131	20	$expr = preg_replace('/(?!- -)[^-a-z_0-9]\\K (?=[^-a-z_0-9])/i', '', $expr);
132
133		// Remove the space between a - and a word character, as long as there's a space before -
134	20	$expr = preg_replace('/ - ([a-z_0-9])/i', ' -$1', $expr);
135
136		// Remove the spaces between a number and a div or "-" operator and the next token
137	20	$expr = preg_replace('/(?:^\|[ \\(])\\d+\\K (div\|-) ?/', '$1', $expr);
138
139		// Remove the space between the div operator the next token
140	20	$expr = preg_replace('/([^-a-z_0-9]div) (?=[$0-9@])/', '$1', $expr);
141
142		// Remove consecutive parentheses where redundant
143	20	$expr = self::removeRedundantParentheses($expr);
144
145		// Restore the literals
146	20	$expr = self::decodeStrings($expr);
147
148	20	return $expr;
149		}
150
151		/**
152		* Remove consecutive parentheses where redundant
153		*
154		* @param string $expr
155		* @return string
156		*/
157	20	protected static function removeRedundantParentheses(string $expr): string
158		{
159	20	preg_match_all('([\\(\\)]\|[^\\(\\)]++)', $expr, $m);
160	20	$tokens = $m[0];
161	20	$tokens[] = '';
162
163	20	$depth = 0;
164	20	$left = [-1 => -1];
165	20	foreach ($tokens as $k => $token)
166		{
167	20	if ($token === '(')
168		{
169	9	$left[$depth] = $k;
170	9	++$depth;
171		}
172	20	elseif ($token === ')')
173		{
174	9	if (--$depth < 0)
175		{
176		throw new RuntimeException("Cannot parse XPath expression '" . $expr . "'");
177		}
178	9	if ($tokens[$k + 1] === ')' && $left[$depth - 1] === $left[$depth] - 1)
179		{
180	3	unset($tokens[$k]);
181	3	unset($tokens[$left[$depth]]);
182		}
183		}
184		}
185
186	20	return implode('', $tokens);
187		}
188
189		/**
190		* Parse an XPath expression that is composed entirely of equality tests between a variable part
191		* and a constant part
192		*
193		* @param string $expr
194		* @return array\|false
195		*/
196	8	public static function parseEqualityExpr($expr)
197		{
198		// Match an equality between a variable and a literal or the concatenation of strings
199		$eq = '(?<equality>'
200		. '(?<key>@[-\\w]+\|\\$\\w+\|\\.)'
201		. '(?<operator>\\s=\\s)'
202		. '(?:'
203		. '(?<literal>(?<string>"[^"]"\|\'[^\']\')\|0\|[1-9][0-9]*)'
204		. '\|'
205		. '(?<concat>concat\\(\\s(?&string)\\s(?:,\\s(?&string)\\s)+\\))'
206		. ')'
207		. '\|'
208		. '(?:(?<literal>(?&literal))\|(?<concat>(?&concat)))(?&operator)(?<key>(?&key))'
209	8	. ')';
210
211		// Match a string that is entirely composed of equality checks separated with "or"
212	8	$regexp = '(^(?J)\\s' . $eq . '\\s(?:or\\s(?&equality)\\s)*$)';
213	8	if (!preg_match($regexp, $expr))
214		{
215	1	return false;
216		}
217
218	7	preg_match_all("((?J)$eq)", $expr, $matches, PREG_SET_ORDER);
219
220	7	$map = [];
221	7	foreach ($matches as $m)
222		{
223	7	$key = $m['key'];
224	7	$value = (!empty($m['concat']))
225	1	? self::evaluateConcat($m['concat'])
226	7	: self::evaluateLiteral($m['literal']);
227
228	7	$map[$key][] = $value;
229		}
230
231	7	return $map;
232		}
233
234		/**
235		* Evaluate a concat() expression where all arguments are string literals
236		*
237		* @param string $expr concat() expression
238		* @return string Expression's value
239		*/
240	1	protected static function evaluateConcat($expr)
241		{
242	1	preg_match_all('(\'[^\']\'\|"[^"]")', $expr, $strings);
243
244	1	$value = '';
245	1	foreach ($strings[0] as $string)
246		{
247	1	$value .= substr($string, 1, -1);
248		}
249
250	1	return $value;
251		}
252
253		/**
254		* Evaluate an XPath literal
255		*
256		* @param string $expr XPath literal
257		* @return string Literal's string value
258		*/
259	6	protected static function evaluateLiteral($expr)
260		{
261	6	if ($expr[0] === '"' \|\| $expr[0] === "'")
262		{
263	6	$expr = substr($expr, 1, -1);
264		}
265
266	6	return $expr;
267		}
268
269		/**
270		* Generate and return a cached XPath parser with a default set of matchers
271		*
272		* @return RecursiveParser
273		*/
274	9	protected static function getXPathParser()
275		{
276	9	static $parser;
277	9	if (!isset($parser))
278		{
279	1	$parser = new RecursiveParser;
280	1	$matchers = [];
281	1	$matchers[] = new BooleanFunctions($parser);
282	1	$matchers[] = new BooleanOperators($parser);
283	1	$matchers[] = new Comparisons($parser);
284	1	$matchers[] = new Core($parser);
285	1	$matchers[] = new Math($parser);
286	1	$matchers[] = new SingleByteStringFunctions($parser);
287
288	1	$parser->setMatchers($matchers);
289		}
290
291	9	return $parser;
292		}
293		}

s9e / TextFormatter

Branch — 2.3/RemoveRedundantParentheses... (567a0c)

XPathHelper::parseEqualityExpr() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like