fluentform / app / Services / Spout / Common / Escaper / XLSX.php

XLSX.php in Fluent Forms – Customizable Contact Forms, Survey, Quiz, & Conversational Form Builder 6.2.5, at app/Services/Spout/Common/Escaper/XLSX.php

186 lines 6.6 KB

Raw Download Zip

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