|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* File containing the {@see AppUtils\URLInfo_Parser} class. |
|
4
|
|
|
* |
|
5
|
|
|
* @package Application Utils |
|
6
|
|
|
* @subpackage URLInfo |
|
7
|
|
|
* @see AppUtils\URLInfo_Parser |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
declare(strict_types=1); |
|
11
|
|
|
|
|
12
|
|
|
namespace AppUtils; |
|
13
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* Handles the URL parsing. |
|
16
|
|
|
* |
|
17
|
|
|
* @package Application Utils |
|
18
|
|
|
* @subpackage URLInfo |
|
19
|
|
|
* @author Sebastian Mordziol <[email protected]> |
|
20
|
|
|
*/ |
|
21
|
|
|
class URLInfo_Parser |
|
22
|
|
|
{ |
|
23
|
|
|
/** |
|
24
|
|
|
* @var string |
|
25
|
|
|
*/ |
|
26
|
|
|
protected $url; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* @var bool |
|
30
|
|
|
*/ |
|
31
|
|
|
protected $isValid = false; |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @var array |
|
35
|
|
|
*/ |
|
36
|
|
|
protected $info; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @var array|NULL |
|
40
|
|
|
*/ |
|
41
|
|
|
protected $error; |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* @var array |
|
45
|
|
|
*/ |
|
46
|
|
|
protected $knownSchemes = array( |
|
47
|
|
|
'ftp', |
|
48
|
|
|
'http', |
|
49
|
|
|
'https', |
|
50
|
|
|
'mailto', |
|
51
|
|
|
'tel', |
|
52
|
|
|
'data', |
|
53
|
|
|
'file' |
|
54
|
|
|
); |
|
55
|
|
|
|
|
56
|
|
|
public function __construct(string $url) |
|
57
|
|
|
{ |
|
58
|
|
|
$this->url = $url; |
|
59
|
|
|
|
|
60
|
|
|
$this->parse(); |
|
61
|
|
|
|
|
62
|
|
|
if(!$this->detectType()) { |
|
63
|
|
|
$this->validate(); |
|
64
|
|
|
} |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
public function getInfo() : array |
|
68
|
|
|
{ |
|
69
|
|
|
return $this->info; |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
protected function parse() |
|
73
|
|
|
{ |
|
74
|
|
|
// fix for parsing unicode characters in URLs: |
|
75
|
|
|
// this is dependent on the machine's locale, |
|
76
|
|
|
// so to ensure this works we temporarily change |
|
77
|
|
|
// it to the always available US UTF8 locale. |
|
78
|
|
|
$prev = setlocale(LC_CTYPE, 'en_US.UTF-8'); |
|
79
|
|
|
|
|
80
|
|
|
$this->info = parse_url($this->url); |
|
81
|
|
|
|
|
82
|
|
|
// restore the previous locale |
|
83
|
|
|
setlocale(LC_CTYPE, $prev); |
|
84
|
|
|
|
|
85
|
|
|
$this->filterParsed(); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
protected function detectType() : bool |
|
89
|
|
|
{ |
|
90
|
|
|
$types = array( |
|
91
|
|
|
'email', |
|
92
|
|
|
'fragmentLink', |
|
93
|
|
|
'phoneLink' |
|
94
|
|
|
); |
|
95
|
|
|
|
|
96
|
|
|
foreach($types as $type) |
|
97
|
|
|
{ |
|
98
|
|
|
$method = 'detectType_'.$type; |
|
99
|
|
|
|
|
100
|
|
|
if($this->$method() === true) |
|
101
|
|
|
{ |
|
102
|
|
|
$this->isValid = true; |
|
103
|
|
|
return true; |
|
104
|
|
|
} |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
return false; |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
protected function validate() |
|
111
|
|
|
{ |
|
112
|
|
|
$validations = array( |
|
113
|
|
|
'schemeIsSet', |
|
114
|
|
|
'schemeIsKnown', |
|
115
|
|
|
'hostIsPresent' |
|
116
|
|
|
); |
|
117
|
|
|
|
|
118
|
|
|
foreach($validations as $validation) |
|
119
|
|
|
{ |
|
120
|
|
|
$method = 'validate_'.$validation; |
|
121
|
|
|
|
|
122
|
|
|
if($this->$method() !== true) { |
|
123
|
|
|
return; |
|
124
|
|
|
} |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
$this->isValid = true; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
protected function validate_hostIsPresent() : bool |
|
131
|
|
|
{ |
|
132
|
|
|
// every link needs a host. This case can happen for ex, if |
|
133
|
|
|
// the link starts with a typo with only one slash, like: |
|
134
|
|
|
// "http:/hostname" |
|
135
|
|
|
if(isset($this->info['host'])) { |
|
136
|
|
|
return true; |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
$this->setError( |
|
140
|
|
|
URLInfo::ERROR_MISSING_HOST, |
|
141
|
|
|
t('Cannot determine the link\'s host name.') . ' ' . |
|
142
|
|
|
t('This usually happens when there\'s a typo somewhere.') |
|
143
|
|
|
); |
|
144
|
|
|
|
|
145
|
|
|
return false; |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
|
|
protected function validate_schemeIsSet() : bool |
|
149
|
|
|
{ |
|
150
|
|
|
if(isset($this->info['scheme'])) { |
|
151
|
|
|
return true; |
|
152
|
|
|
} |
|
153
|
|
|
|
|
154
|
|
|
// no scheme found: it may be an email address without the mailto: |
|
155
|
|
|
// It can't be a variable, since without the scheme it would already |
|
156
|
|
|
// have been recognized as a vaiable only link. |
|
157
|
|
|
$this->setError( |
|
158
|
|
|
URLInfo::ERROR_MISSING_SCHEME, |
|
159
|
|
|
t('Cannot determine the link\'s scheme, e.g. %1$s.', 'http') |
|
160
|
|
|
); |
|
161
|
|
|
|
|
162
|
|
|
return false; |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
protected function validate_schemeIsKnown() : bool |
|
166
|
|
|
{ |
|
167
|
|
|
if(in_array($this->info['scheme'], $this->knownSchemes)) { |
|
168
|
|
|
return true; |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
|
|
$this->setError( |
|
172
|
|
|
URLInfo::ERROR_INVALID_SCHEME, |
|
173
|
|
|
t('The scheme %1$s is not supported for links.', $this->info['scheme']) . ' ' . |
|
174
|
|
|
t('Valid schemes are: %1$s.', implode(', ', $this->knownSchemes)) |
|
175
|
|
|
); |
|
176
|
|
|
|
|
177
|
|
|
return false; |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
/** |
|
181
|
|
|
* Goes through all information in the parse_url result |
|
182
|
|
|
* array, and attempts to fix any user errors in formatting |
|
183
|
|
|
* that can be recovered from, mostly regarding stray spaces. |
|
184
|
|
|
*/ |
|
185
|
|
|
protected function filterParsed() |
|
186
|
|
|
{ |
|
187
|
|
|
foreach($this->info as $key => $val) |
|
188
|
|
|
{ |
|
189
|
|
|
if(is_string($val)) { |
|
190
|
|
|
$this->info[$key] = trim($val); |
|
191
|
|
|
} |
|
192
|
|
|
} |
|
193
|
|
|
|
|
194
|
|
|
$this->info['params'] = array(); |
|
195
|
|
|
$this->info['type'] = URLInfo::TYPE_URL; |
|
196
|
|
|
|
|
197
|
|
|
if(isset($this->info['host'])) { |
|
198
|
|
|
$this->info['host'] = str_replace(' ', '', $this->info['host']); |
|
199
|
|
|
} |
|
200
|
|
|
|
|
201
|
|
|
if(isset($this->info['path'])) { |
|
202
|
|
|
$this->info['path'] = str_replace(' ', '', $this->info['path']); |
|
203
|
|
|
} |
|
204
|
|
|
|
|
205
|
|
|
if(isset($this->info['query']) && !empty($this->info['query'])) |
|
206
|
|
|
{ |
|
207
|
|
|
$this->info['params'] = \AppUtils\ConvertHelper::parseQueryString($this->info['query']); |
|
208
|
|
|
ksort($this->info['params']); |
|
209
|
|
|
} |
|
210
|
|
|
} |
|
211
|
|
|
|
|
212
|
|
|
protected function detectType_email() : bool |
|
213
|
|
|
{ |
|
214
|
|
|
if(isset($this->info['scheme']) && $this->info['scheme'] == 'mailto') { |
|
215
|
|
|
$this->info['type'] = URLInfo::TYPE_EMAIL; |
|
216
|
|
|
return true; |
|
217
|
|
|
} |
|
218
|
|
|
|
|
219
|
|
|
if(isset($this->info['path']) && preg_match(\AppUtils\RegexHelper::REGEX_EMAIL, $this->info['path'])) |
|
220
|
|
|
{ |
|
221
|
|
|
$this->info['scheme'] = 'mailto'; |
|
222
|
|
|
$this->info['type'] = URLInfo::TYPE_EMAIL; |
|
223
|
|
|
return true; |
|
224
|
|
|
} |
|
225
|
|
|
|
|
226
|
|
|
return false; |
|
227
|
|
|
} |
|
228
|
|
|
|
|
229
|
|
|
protected function detectType_fragmentLink() : bool |
|
230
|
|
|
{ |
|
231
|
|
|
if(isset($this->info['fragment']) && !isset($this->info['scheme'])) { |
|
232
|
|
|
$this->info['type'] = URLInfo::TYPE_FRAGMENT; |
|
233
|
|
|
return true; |
|
234
|
|
|
} |
|
235
|
|
|
|
|
236
|
|
|
return false; |
|
237
|
|
|
} |
|
238
|
|
|
|
|
239
|
|
|
protected function detectType_phoneLink() : bool |
|
240
|
|
|
{ |
|
241
|
|
|
if(isset($this->info['scheme']) && $this->info['scheme'] == 'tel') { |
|
242
|
|
|
$this->info['type'] = URLInfo::TYPE_PHONE; |
|
243
|
|
|
return true; |
|
244
|
|
|
} |
|
245
|
|
|
|
|
246
|
|
|
return false; |
|
247
|
|
|
} |
|
248
|
|
|
|
|
249
|
|
|
protected function setError(int $code, string $message) |
|
250
|
|
|
{ |
|
251
|
|
|
$this->isValid = false; |
|
252
|
|
|
|
|
253
|
|
|
$this->error = array( |
|
254
|
|
|
'code' => $code, |
|
255
|
|
|
'message' => $message |
|
256
|
|
|
); |
|
257
|
|
|
} |
|
258
|
|
|
|
|
259
|
|
|
public function isValid() : bool |
|
260
|
|
|
{ |
|
261
|
|
|
return $this->isValid; |
|
262
|
|
|
} |
|
263
|
|
|
|
|
264
|
|
|
public function getErrorMessage() : string |
|
265
|
|
|
{ |
|
266
|
|
|
if(isset($this->error)) { |
|
267
|
|
|
return $this->error['message']; |
|
268
|
|
|
} |
|
269
|
|
|
|
|
270
|
|
|
return ''; |
|
271
|
|
|
} |
|
272
|
|
|
|
|
273
|
|
|
public function getErrorCode() : int |
|
274
|
|
|
{ |
|
275
|
|
|
if(isset($this->error)) { |
|
276
|
|
|
return $this->error['code']; |
|
277
|
|
|
} |
|
278
|
|
|
|
|
279
|
|
|
return -1; |
|
280
|
|
|
} |
|
281
|
|
|
} |
|
282
|
|
|
|