XLSX::unescape() - Code Metrics - Inspection of "Improve XLSX Escaper performance (#305)" - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 5e7a17...435a9a )

by Adrien

created 2016-09-03 18:04 UTC

XLSX::unescape() A

↳ Parent: XLSX

Complexity

Conditions	1
Paths	1

Size

Total Lines	7
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	4
CRAP Score	1

Importance

Changes

Metric	Value
dl	0
loc	7
rs	9.4285
c	0
b	0
f	0
ccs	4
cts	4
cp	1
cc	1
eloc	4
nc	1
nop	1
crap	1

<?php

namespace Box\Spout\Common\Escaper;

use Box\Spout\Common\Singleton;

/**
 * Class XLSX
 * Provides functions to escape and unescape data for XLSX files
 *
 * @package Box\Spout\Common\Escaper
 */
class XLSX implements EscaperInterface
{
    use Singleton;

    /** @var string Regex pattern to detect control characters that need to be escaped */
    protected $escapableControlCharactersPattern;

    /** @var string[] Map containing control characters to be escaped (key) and their escaped value (value) */
    protected $controlCharactersEscapingMap;

    /** @var string[] Map containing control characters to be escaped (value) and their escaped value (key) */
    protected $controlCharactersEscapingReverseMap;

    /**
     * Initializes the singleton instance
     */

    protected function init()
    {
        $this->escapableControlCharactersPattern = $this->getEscapableControlCharactersPattern();
        $this->controlCharactersEscapingMap = $this->getControlCharactersEscapingMap();
        $this->controlCharactersEscapingReverseMap = array_flip($this->controlCharactersEscapingMap);
    }

    /**
     * Escapes the given string to make it compatible with XLSX
     *
     * @param string $string The string to escape
     * @return string The escaped string
     */
    public function escape($string)
    {
        $escapedString = $this->escapeControlCharacters($string);
        $escapedString = htmlspecialchars($escapedString, ENT_QUOTES);

        return $escapedString;
    }

    /**
     * Unescapes the given string to make it compatible with XLSX
     *
     * @param string $string The string to unescape
     * @return string The unescaped string
     */
    public function unescape($string)
    {
        $unescapedString = htmlspecialchars_decode($string, ENT_QUOTES);
        $unescapedString = $this->unescapeControlCharacters($unescapedString);

        return $unescapedString;
    }

    /**
     * @return string Regex pattern containing all escapable control characters
     */
    protected function getEscapableControlCharactersPattern()
    {
        // control characters values are from 0 to 1F (hex values) in the ASCII table
        // some characters should not be escaped though: "\t", "\r" and "\n".
        return '[\x00-\x08' .
                // skipping "\t" (0x9) and "\n" (0xA)
                '\x0B-\x0C' .
                // skipping "\r" (0xD)
                '\x0E-\x1F]';
    }

    /**
     * Builds the map containing control characters to be escaped
     * mapped to their escaped values.
     * "\t", "\r" and "\n" don't need to be escaped.
     *
     * NOTE: the logic has been adapted from the XlsxWriter library (BSD License)
     * @link https://github.com/jmcnamara/XlsxWriter/blob/f1e610f29/xlsxwriter/sharedstrings.py#L89
     *
     * @return string[]
     */
    protected function getControlCharactersEscapingMap()
    {
        $controlCharactersEscapingMap = [];

        // control characters values are from 0 to 1F (hex values) in the ASCII table
        for ($charValue = 0x00; $charValue <= 0x1F; $charValue++) {
            $character = chr($charValue);
            if (preg_match("/{$this->escapableControlCharactersPattern}/", $character)) {
                $charHexValue = dechex($charValue);
                $escapedChar = '_x' . sprintf('%04s' , strtoupper($charHexValue)) . '_';
                $controlCharactersEscapingMap[$escapedChar] = $character;
            }
        }

        return $controlCharactersEscapingMap;
    }

    /**
     * Converts PHP control characters from the given string to OpenXML escaped control characters
     *
     * Excel escapes control characters with _xHHHH_ and also escapes any
     * literal strings of that type by encoding the leading underscore.
     * So "\0" -> _x0000_ and "_x0000_" -> _x005F_x0000_.
     *
     * NOTE: the logic has been adapted from the XlsxWriter library (BSD License)
     * @link https://github.com/jmcnamara/XlsxWriter/blob/f1e610f29/xlsxwriter/sharedstrings.py#L89
     *
     * @param string $string String to escape
     * @return string
     */
    protected function escapeControlCharacters($string)
    {
        $escapedString = $this->escapeEscapeCharacter($string);

        // if no control characters
        if (!preg_match("/{$this->escapableControlCharactersPattern}/", $escapedString)) {
            return $escapedString;
        }

        return preg_replace_callback("/({$this->escapableControlCharactersPattern})/", function($matches) {
            return $this->controlCharactersEscapingReverseMap[$matches[0]];
        }, $escapedString);
    }

    /**
     * Escapes the escape character: "_x0000_" -> "_x005F_x0000_"
     *
     * @param string $string String to escape
     * @return string The escaped string
     */
    protected function escapeEscapeCharacter($string)
    {
        return preg_replace('/_(x[\dA-F]{4})_/', '_x005F_$1_', $string);
    }

    /**
     * Converts OpenXML escaped control characters from the given string to PHP control characters
     *
     * Excel escapes control characters with _xHHHH_ and also escapes any
     * literal strings of that type by encoding the leading underscore.
     * So "_x0000_" -> "\0" and "_x005F_x0000_" -> "_x0000_"
     *
     * NOTE: the logic has been adapted from the XlsxWriter library (BSD License)
     * @link https://github.com/jmcnamara/XlsxWriter/blob/f1e610f29/xlsxwriter/sharedstrings.py#L89
     *
     * @param string $string String to unescape
     * @return string
     */
    protected function unescapeControlCharacters($string)
    {
        $unescapedString = $string;

        foreach ($this->controlCharactersEscapingMap as $escapedCharValue => $charValue) {
            // only unescape characters that don't contain the escaped escape character for now
            $unescapedString = preg_replace("/(?<!_x005F)($escapedCharValue)/", $charValue, $unescapedString);
        }

        return $this->unescapeEscapeCharacter($unescapedString);
    }

    /**
     * Unecapes the escape character: "_x005F_x0000_" => "_x0000_"
     *
     * @param string $string String to unescape
     * @return string The unescaped string
     */
    protected function unescapeEscapeCharacter($string)
    {
        return preg_replace('/_x005F(_x[\dA-F]{4}_)/', '$1', $string);
    }
}


1		<?php
2
3		namespace Box\Spout\Common\Escaper;
4
5		use Box\Spout\Common\Singleton;
6
7		/**
8		* Class XLSX
9		* Provides functions to escape and unescape data for XLSX files
10		*
11		* @package Box\Spout\Common\Escaper
12		*/
13		class XLSX implements EscaperInterface
14		{
15		use Singleton;
16
17		/** @var string Regex pattern to detect control characters that need to be escaped */
18		protected $escapableControlCharactersPattern;
19
20		/** @var string[] Map containing control characters to be escaped (key) and their escaped value (value) */
21		protected $controlCharactersEscapingMap;
22
23		/** @var string[] Map containing control characters to be escaped (value) and their escaped value (key) */
24		protected $controlCharactersEscapingReverseMap;
25
26		/**
27		* Initializes the singleton instance
28		*/
		1 ignored issue – show introduced 2016-06-13 21:35 UTC by Report Bug Copy Issue Report Instead of declaring the constructor as final, maybe you should declare the entire class as final. Loading history...
29	3	protected function init()
30		{
31	3	$this->escapableControlCharactersPattern = $this->getEscapableControlCharactersPattern();
32	3	$this->controlCharactersEscapingMap = $this->getControlCharactersEscapingMap();
33	3	$this->controlCharactersEscapingReverseMap = array_flip($this->controlCharactersEscapingMap);
34	3	}
35
36		/**
37		* Escapes the given string to make it compatible with XLSX
38		*
39		* @param string $string The string to escape
40		* @return string The escaped string
41		*/
42	117	public function escape($string)
43		{
44	117	$escapedString = $this->escapeControlCharacters($string);
45	117	$escapedString = htmlspecialchars($escapedString, ENT_QUOTES);
46
47	117	return $escapedString;
48		}
49
50		/**
51		* Unescapes the given string to make it compatible with XLSX
52		*
53		* @param string $string The string to unescape
54		* @return string The unescaped string
55		*/
56	150	public function unescape($string)
57		{
58	150	$unescapedString = htmlspecialchars_decode($string, ENT_QUOTES);
59	150	$unescapedString = $this->unescapeControlCharacters($unescapedString);
60
61	150	return $unescapedString;
62		}
63
64		/**
65		* @return string Regex pattern containing all escapable control characters
66		*/
67	3	protected function getEscapableControlCharactersPattern()
68		{
69		// control characters values are from 0 to 1F (hex values) in the ASCII table
70		// some characters should not be escaped though: "\t", "\r" and "\n".
71		return '[\x00-\x08' .
72		// skipping "\t" (0x9) and "\n" (0xA)
73	3	'\x0B-\x0C' .
74		// skipping "\r" (0xD)
75	3	'\x0E-\x1F]';
76		}
77
78		/**
79		* Builds the map containing control characters to be escaped
80		* mapped to their escaped values.
81		* "\t", "\r" and "\n" don't need to be escaped.
82		*
83		* NOTE: the logic has been adapted from the XlsxWriter library (BSD License)
84		* @link https://github.com/jmcnamara/XlsxWriter/blob/f1e610f29/xlsxwriter/sharedstrings.py#L89
85		*
86		* @return string[]
87		*/
88	3	protected function getControlCharactersEscapingMap()
89		{
90	3	$controlCharactersEscapingMap = [];
91
92		// control characters values are from 0 to 1F (hex values) in the ASCII table
93	3	for ($charValue = 0x00; $charValue <= 0x1F; $charValue++) {
94	3	$character = chr($charValue);
95	3	if (preg_match("/{$this->escapableControlCharactersPattern}/", $character)) {
96	3	$charHexValue = dechex($charValue);
97	3	$escapedChar = '_x' . sprintf('%04s' , strtoupper($charHexValue)) . '_';
98	3	$controlCharactersEscapingMap[$escapedChar] = $character;
99	3	}
100	3	}
101
102	3	return $controlCharactersEscapingMap;
103		}
104
105		/**
106		* Converts PHP control characters from the given string to OpenXML escaped control characters
107		*
108		* Excel escapes control characters with _xHHHH_ and also escapes any
109		* literal strings of that type by encoding the leading underscore.
110		* So "\0" -> _x0000_ and "_x0000_" -> _x005F_x0000_.
111		*
112		* NOTE: the logic has been adapted from the XlsxWriter library (BSD License)
113		* @link https://github.com/jmcnamara/XlsxWriter/blob/f1e610f29/xlsxwriter/sharedstrings.py#L89
114		*
115		* @param string $string String to escape
116		* @return string
117		*/
118	117	protected function escapeControlCharacters($string)
119		{
120	117	$escapedString = $this->escapeEscapeCharacter($string);
121
122		// if no control characters
123	117	if (!preg_match("/{$this->escapableControlCharactersPattern}/", $escapedString)) {
124	102	return $escapedString;
125		}
126
127	18	return preg_replace_callback("/({$this->escapableControlCharactersPattern})/", function($matches) {
128	18	return $this->controlCharactersEscapingReverseMap[$matches[0]];
129	18	}, $escapedString);
130		}
131
132		/**
133		* Escapes the escape character: "_x0000_" -> "_x005F_x0000_"
134		*
135		* @param string $string String to escape
136		* @return string The escaped string
137		*/
138	117	protected function escapeEscapeCharacter($string)
139		{
140	117	return preg_replace('/_(x[\dA-F]{4})_/', '_x005F_$1_', $string);
141		}
142
143		/**
144		* Converts OpenXML escaped control characters from the given string to PHP control characters
145		*
146		* Excel escapes control characters with _xHHHH_ and also escapes any
147		* literal strings of that type by encoding the leading underscore.
148		* So "_x0000_" -> "\0" and "_x005F_x0000_" -> "_x0000_"
149		*
150		* NOTE: the logic has been adapted from the XlsxWriter library (BSD License)
151		* @link https://github.com/jmcnamara/XlsxWriter/blob/f1e610f29/xlsxwriter/sharedstrings.py#L89
152		*
153		* @param string $string String to unescape
154		* @return string
155		*/
156	150	protected function unescapeControlCharacters($string)
157		{
158	150	$unescapedString = $string;
159
160	150	foreach ($this->controlCharactersEscapingMap as $escapedCharValue => $charValue) {
161		// only unescape characters that don't contain the escaped escape character for now
162	150	$unescapedString = preg_replace("/(?<!_x005F)($escapedCharValue)/", $charValue, $unescapedString);
163	150	}
164
165	150	return $this->unescapeEscapeCharacter($unescapedString);
166		}
167
168		/**
169		* Unecapes the escape character: "_x005F_x0000_" => "_x0000_"
170		*
171		* @param string $string String to unescape
172		* @return string The unescaped string
173		*/
174	150	protected function unescapeEscapeCharacter($string)
175		{
176	150	return preg_replace('/_x005F(_x[\dA-F]{4}_)/', '$1', $string);
177		}
178		}
179

box / spout

Push — master ( 5e7a17...435a9a )

XLSX::unescape() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like