XLSX::unescape() - Code Metrics - Inspection of "Refactor readers for better di" - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — develop_3.0 (#457)

by Adrien

created 2017-08-26 21:44 UTC

XLSX::unescape() A

↳ Parent: XLSX

Complexity

Conditions	1
Paths	1

Size

Total Lines	14
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	4
CRAP Score	1

Importance

Changes

Metric	Value
dl	0
loc	14
ccs	4
cts	4
cp	1
rs	9.4285
c	0
b	0
f	0
cc	1
eloc	4
nc	1
nop	1
crap	1

<?php

namespace Box\Spout\Common\Helper\Escaper;

/**
 * Class XLSX
 * Provides functions to escape and unescape data for XLSX files
 *
 * @package Box\Spout\Common\Helper\Escaper
 */
class XLSX implements EscaperInterface
{
    /** @var bool Whether the escaper has already been initialized */
    private $isAlreadyInitialized = false;

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

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

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

    /**
     * Initializes the control characters if not already done
     */
    protected function initIfNeeded()
    {
        if (!$this->isAlreadyInitialized) {
            $this->escapableControlCharactersPattern = $this->getEscapableControlCharactersPattern();
            $this->controlCharactersEscapingMap = $this->getControlCharactersEscapingMap();
            $this->controlCharactersEscapingReverseMap = array_flip($this->controlCharactersEscapingMap);

            $this->isAlreadyInitialized = true;
        }
    }

    /**
     * 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)
    {
        $this->initIfNeeded();

        $escapedString = $this->escapeControlCharacters($string);
        // @NOTE: Using ENT_NOQUOTES as only XML entities ('<', '>', '&') need to be encoded.
        //        Single and double quotes can be left as is.
        $escapedString = htmlspecialchars($escapedString, ENT_NOQUOTES);

        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)
    {
        $this->initIfNeeded();

        // ==============
        // =   WARNING  =
        // ==============
        // It is assumed that the given string has already had its XML entities decoded.
        // This is true if the string is coming from a DOMNode (as DOMNode already decode XML entities on creation).
        // Therefore there is no need to call "htmlspecialchars_decode()".
        $unescapedString = $this->unescapeControlCharacters($string);

        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\Helper\Escaper;
4
5		/**
6		* Class XLSX
7		* Provides functions to escape and unescape data for XLSX files
8		*
9		* @package Box\Spout\Common\Helper\Escaper
10		*/
11		class XLSX implements EscaperInterface
12		{
13		/** @var bool Whether the escaper has already been initialized */
14		private $isAlreadyInitialized = false;
15
16		/** @var string Regex pattern to detect control characters that need to be escaped */
17		private $escapableControlCharactersPattern;
18
19		/** @var string[] Map containing control characters to be escaped (key) and their escaped value (value) */
20		private $controlCharactersEscapingMap;
21
22		/** @var string[] Map containing control characters to be escaped (value) and their escaped value (key) */
23		private $controlCharactersEscapingReverseMap;
24
25		/**
26		* Initializes the control characters if not already done
27		*/
28	95	protected function initIfNeeded()
29		{
30	95	if (!$this->isAlreadyInitialized) {
31	95	$this->escapableControlCharactersPattern = $this->getEscapableControlCharactersPattern();
32	95	$this->controlCharactersEscapingMap = $this->getControlCharactersEscapingMap();
33	95	$this->controlCharactersEscapingReverseMap = array_flip($this->controlCharactersEscapingMap);
34
35	95	$this->isAlreadyInitialized = true;
36		}
37	95	}
38
39		/**
40		* Escapes the given string to make it compatible with XLSX
41		*
42		* @param string $string The string to escape
43		* @return string The escaped string
44		*/
45	47	public function escape($string)
46		{
47	47	$this->initIfNeeded();
48
49	47	$escapedString = $this->escapeControlCharacters($string);
50		// @NOTE: Using ENT_NOQUOTES as only XML entities ('<', '>', '&') need to be encoded.
51		// Single and double quotes can be left as is.
52	47	$escapedString = htmlspecialchars($escapedString, ENT_NOQUOTES);
53
54	47	return $escapedString;
55		}
56
57		/**
58		* Unescapes the given string to make it compatible with XLSX
59		*
60		* @param string $string The string to unescape
61		* @return string The unescaped string
62		*/
63	48	public function unescape($string)
64		{
65	48	$this->initIfNeeded();
66
67		// ==============
68		// = WARNING =
69		// ==============
70		// It is assumed that the given string has already had its XML entities decoded.
71		// This is true if the string is coming from a DOMNode (as DOMNode already decode XML entities on creation).
72		// Therefore there is no need to call "htmlspecialchars_decode()".
73	48	$unescapedString = $this->unescapeControlCharacters($string);
74
75	48	return $unescapedString;
76		}
77
78		/**
79		* @return string Regex pattern containing all escapable control characters
80		*/
81	95	protected function getEscapableControlCharactersPattern()
82		{
83		// control characters values are from 0 to 1F (hex values) in the ASCII table
84		// some characters should not be escaped though: "\t", "\r" and "\n".
85		return '[\x00-\x08' .
86		// skipping "\t" (0x9) and "\n" (0xA)
87		'\x0B-\x0C' .
88		// skipping "\r" (0xD)
89	95	'\x0E-\x1F]';
90		}
91
92		/**
93		* Builds the map containing control characters to be escaped
94		* mapped to their escaped values.
95		* "\t", "\r" and "\n" don't need to be escaped.
96		*
97		* NOTE: the logic has been adapted from the XlsxWriter library (BSD License)
98		* @link https://github.com/jmcnamara/XlsxWriter/blob/f1e610f29/xlsxwriter/sharedstrings.py#L89
99		*
100		* @return string[]
101		*/
102	95	protected function getControlCharactersEscapingMap()
103		{
104	95	$controlCharactersEscapingMap = [];
105
106		// control characters values are from 0 to 1F (hex values) in the ASCII table
107	95	for ($charValue = 0x00; $charValue <= 0x1F; $charValue++) {
108	95	$character = chr($charValue);
109	95	if (preg_match("/{$this->escapableControlCharactersPattern}/", $character)) {
110	95	$charHexValue = dechex($charValue);
111	95	$escapedChar = '_x' . sprintf('%04s' , strtoupper($charHexValue)) . '_';
112	95	$controlCharactersEscapingMap[$escapedChar] = $character;
113		}
114		}
115
116	95	return $controlCharactersEscapingMap;
117		}
118
119		/**
120		* Converts PHP control characters from the given string to OpenXML escaped control characters
121		*
122		* Excel escapes control characters with _xHHHH_ and also escapes any
123		* literal strings of that type by encoding the leading underscore.
124		* So "\0" -> _x0000_ and "_x0000_" -> _x005F_x0000_.
125		*
126		* NOTE: the logic has been adapted from the XlsxWriter library (BSD License)
127		* @link https://github.com/jmcnamara/XlsxWriter/blob/f1e610f29/xlsxwriter/sharedstrings.py#L89
128		*
129		* @param string $string String to escape
130		* @return string
131		*/
132	47	protected function escapeControlCharacters($string)
133		{
134	47	$escapedString = $this->escapeEscapeCharacter($string);
135
136		// if no control characters
137	47	if (!preg_match("/{$this->escapableControlCharactersPattern}/", $escapedString)) {
138	42	return $escapedString;
139		}
140
141	6	return preg_replace_callback("/({$this->escapableControlCharactersPattern})/", function($matches) {
142	6	return $this->controlCharactersEscapingReverseMap[$matches[0]];
143	6	}, $escapedString);
144		}
145
146		/**
147		* Escapes the escape character: "_x0000_" -> "_x005F_x0000_"
148		*
149		* @param string $string String to escape
150		* @return string The escaped string
151		*/
152	47	protected function escapeEscapeCharacter($string)
153		{
154	47	return preg_replace('/_(x[\dA-F]{4})_/', '_x005F_$1_', $string);
155		}
156
157		/**
158		* Converts OpenXML escaped control characters from the given string to PHP control characters
159		*
160		* Excel escapes control characters with _xHHHH_ and also escapes any
161		* literal strings of that type by encoding the leading underscore.
162		* So "_x0000_" -> "\0" and "_x005F_x0000_" -> "_x0000_"
163		*
164		* NOTE: the logic has been adapted from the XlsxWriter library (BSD License)
165		* @link https://github.com/jmcnamara/XlsxWriter/blob/f1e610f29/xlsxwriter/sharedstrings.py#L89
166		*
167		* @param string $string String to unescape
168		* @return string
169		*/
170	48	protected function unescapeControlCharacters($string)
171		{
172	48	$unescapedString = $string;
173
174	48	foreach ($this->controlCharactersEscapingMap as $escapedCharValue => $charValue) {
175		// only unescape characters that don't contain the escaped escape character for now
176	48	$unescapedString = preg_replace("/(?<!_x005F)($escapedCharValue)/", $charValue, $unescapedString);
177		}
178
179	48	return $this->unescapeEscapeCharacter($unescapedString);
180		}
181
182		/**
183		* Unecapes the escape character: "_x005F_x0000_" => "_x0000_"
184		*
185		* @param string $string String to unescape
186		* @return string The unescaped string
187		*/
188	48	protected function unescapeEscapeCharacter($string)
189		{
190	48	return preg_replace('/_x005F(_x[\dA-F]{4}_)/', '$1', $string);
191		}
192		}
193

box / spout

Pull Request — develop_3.0 (#457)

XLSX::unescape() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like