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; |
9
|
|
|
|
10
|
|
|
use DOMDocument; |
11
|
|
|
use InvalidArgumentException; |
12
|
|
|
|
13
|
|
|
abstract class Renderer |
14
|
|
|
{ |
15
|
|
|
/** |
16
|
|
|
* @var array Associative array of [paramName => paramValue] |
17
|
|
|
*/ |
18
|
|
|
protected $params = []; |
19
|
|
|
|
20
|
|
|
/** |
21
|
|
|
* @var mixed Saved locale |
22
|
|
|
*/ |
23
|
|
|
protected $savedLocale = 0; |
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* Create a return a new DOMDocument loaded with given XML |
27
|
|
|
* |
28
|
|
|
* @param string $xml Source XML |
29
|
|
|
* @return DOMDocument |
30
|
|
|
*/ |
31
|
29 |
|
protected function loadXML($xml) |
32
|
|
|
{ |
33
|
29 |
|
$this->checkUnsupported($xml); |
34
|
|
|
|
35
|
|
|
// Activate small nodes allocation and relax LibXML's hardcoded limits if applicable. Limits |
36
|
|
|
// on tags can be set during configuration |
37
|
21 |
|
$flags = (LIBXML_VERSION >= 20700) ? LIBXML_COMPACT | LIBXML_PARSEHUGE : 0; |
38
|
|
|
|
39
|
21 |
|
$useErrors = libxml_use_internal_errors(true); |
40
|
21 |
|
$dom = new DOMDocument; |
41
|
21 |
|
$success = $dom->loadXML($xml, $flags); |
42
|
21 |
|
libxml_use_internal_errors($useErrors); |
43
|
|
|
|
44
|
21 |
|
if (!$success) |
45
|
|
|
{ |
46
|
3 |
|
throw new InvalidArgumentException('Cannot load XML: ' . libxml_get_last_error()->message); |
47
|
|
|
} |
48
|
|
|
|
49
|
18 |
|
return $dom; |
50
|
|
|
} |
51
|
|
|
|
52
|
|
|
/** |
53
|
|
|
* Render an intermediate representation |
54
|
|
|
* |
55
|
|
|
* @param string $xml Intermediate representation |
56
|
|
|
* @return string Rendered result |
57
|
|
|
*/ |
58
|
46 |
|
public function render($xml) |
59
|
|
|
{ |
60
|
46 |
|
if (substr($xml, 0, 3) === '<t>' && substr($xml, -4) === '</t>') |
61
|
|
|
{ |
62
|
10 |
|
return $this->renderPlainText($xml); |
63
|
|
|
} |
64
|
|
|
else |
65
|
|
|
{ |
66
|
36 |
|
return $this->renderRichText(preg_replace('(<[eis]>[^<]*</[eis]>)', '', $xml)); |
67
|
|
|
} |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
/** |
71
|
|
|
* Render an intermediate representation of plain text |
72
|
|
|
* |
73
|
|
|
* @param string $xml Intermediate representation |
74
|
|
|
* @return string Rendered result |
75
|
|
|
*/ |
76
|
10 |
|
protected function renderPlainText($xml) |
77
|
|
|
{ |
78
|
|
|
// Remove the <t> and </t> tags |
79
|
10 |
|
$html = substr($xml, 3, -4); |
80
|
|
|
|
81
|
|
|
// Replace all <br/> with <br> |
82
|
10 |
|
$html = str_replace('<br/>', '<br>', $html); |
83
|
|
|
|
84
|
|
|
// Decode encoded characters from the Supplementary Multilingual Plane |
85
|
10 |
|
$html = $this->decodeSMP($html); |
86
|
|
|
|
87
|
10 |
|
return $html; |
88
|
|
|
} |
89
|
|
|
|
90
|
|
|
/** |
91
|
|
|
* Render an intermediate representation of rich text |
92
|
|
|
* |
93
|
|
|
* @param string $xml Intermediate representation |
94
|
|
|
* @return string Rendered result |
95
|
|
|
*/ |
96
|
|
|
abstract protected function renderRichText($xml); |
97
|
|
|
|
98
|
|
|
/** |
99
|
|
|
* Get the value of a parameter |
100
|
|
|
* |
101
|
|
|
* @param string $paramName |
102
|
|
|
* @return string |
103
|
|
|
*/ |
104
|
6 |
|
public function getParameter($paramName) |
105
|
|
|
{ |
106
|
6 |
|
return (isset($this->params[$paramName])) ? $this->params[$paramName] : ''; |
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
/** |
110
|
|
|
* Get the values of all parameters |
111
|
|
|
* |
112
|
|
|
* @return array Associative array of parameter names and values |
113
|
|
|
*/ |
114
|
5 |
|
public function getParameters() |
115
|
|
|
{ |
116
|
5 |
|
return $this->params; |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
/** |
120
|
|
|
* Set the value of a parameter from the stylesheet |
121
|
|
|
* |
122
|
|
|
* @param string $paramName Parameter name |
123
|
|
|
* @param mixed $paramValue Parameter's value |
124
|
|
|
* @return void |
125
|
|
|
*/ |
126
|
6 |
|
public function setParameter($paramName, $paramValue) |
127
|
|
|
{ |
128
|
6 |
|
$this->params[$paramName] = (string) $paramValue; |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
/** |
132
|
|
|
* Set the values of several parameters from the stylesheet |
133
|
|
|
* |
134
|
|
|
* @param string $params Associative array of [parameter name => parameter value] |
135
|
|
|
* @return void |
136
|
|
|
*/ |
137
|
3 |
|
public function setParameters(array $params) |
138
|
|
|
{ |
139
|
3 |
|
foreach ($params as $paramName => $paramValue) |
140
|
|
|
{ |
141
|
3 |
|
$this->setParameter($paramName, $paramValue); |
142
|
|
|
} |
143
|
|
|
} |
144
|
|
|
|
145
|
|
|
/** |
146
|
|
|
* Test for the presence of unsupported XML and throw an exception if found |
147
|
|
|
* |
148
|
|
|
* @param string $xml XML |
149
|
|
|
* @return void |
150
|
|
|
*/ |
151
|
29 |
|
protected function checkUnsupported($xml) |
152
|
|
|
{ |
153
|
29 |
|
if (preg_match('((?<=<)[!?])', $xml, $m)) |
154
|
|
|
{ |
155
|
|
|
$errors = [ |
156
|
8 |
|
'!' => 'DTDs, CDATA nodes and comments are not allowed', |
157
|
|
|
'?' => 'Processing instructions are not allowed' |
158
|
|
|
]; |
159
|
|
|
|
160
|
8 |
|
throw new InvalidArgumentException($errors[$m[0]]); |
161
|
|
|
} |
162
|
|
|
} |
163
|
|
|
|
164
|
|
|
/** |
165
|
|
|
* Decode encoded characters from the Supplementary Multilingual Plane |
166
|
|
|
* |
167
|
|
|
* @param string $str Encoded string |
168
|
|
|
* @return string Decoded string |
169
|
|
|
*/ |
170
|
21 |
|
protected function decodeSMP($str) |
171
|
|
|
{ |
172
|
21 |
|
if (strpos($str, '&#') === false) |
173
|
|
|
{ |
174
|
11 |
|
return $str; |
175
|
|
|
} |
176
|
|
|
|
177
|
10 |
|
return preg_replace_callback('(&#(?:x[0-9A-Fa-f]+|[0-9]+);)', __CLASS__ . '::decodeEntity', $str); |
178
|
|
|
} |
179
|
|
|
|
180
|
|
|
/** |
181
|
|
|
* Decode a matched SGML entity |
182
|
|
|
* |
183
|
|
|
* @param string[] $m Captures from PCRE |
184
|
|
|
* @return string Decoded entity |
185
|
|
|
*/ |
186
|
10 |
|
protected static function decodeEntity(array $m) |
187
|
|
|
{ |
188
|
10 |
|
return htmlspecialchars(html_entity_decode($m[0], ENT_QUOTES, 'UTF-8'), ENT_COMPAT); |
189
|
|
|
} |
190
|
|
|
|
191
|
|
|
/** |
192
|
|
|
* Restore the original locale |
193
|
|
|
*/ |
194
|
25 |
|
protected function restoreLocale(): void |
195
|
|
|
{ |
196
|
25 |
|
if ($this->savedLocale !== 'C') |
197
|
|
|
{ |
198
|
14 |
|
setlocale(LC_NUMERIC, $this->savedLocale); |
199
|
|
|
} |
200
|
|
|
} |
201
|
|
|
|
202
|
|
|
/** |
203
|
|
|
* Temporarily set the locale to C |
204
|
|
|
*/ |
205
|
30 |
|
protected function setLocale(): void |
206
|
|
|
{ |
207
|
30 |
|
$this->savedLocale = setlocale(LC_NUMERIC, 0); |
208
|
30 |
|
if ($this->savedLocale !== 'C') |
209
|
|
|
{ |
210
|
16 |
|
setlocale(LC_NUMERIC, 'C'); |
211
|
|
|
} |
212
|
|
|
} |
213
|
|
|
} |