1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
|
4
|
|
|
namespace seosazi\PathConverter; |
5
|
|
|
|
6
|
|
|
|
7
|
|
|
/** |
8
|
|
|
* Convert paths relative with base tag and domain to absolute path |
9
|
|
|
* |
10
|
|
|
* E.g. |
11
|
|
|
* href="style.css" |
12
|
|
|
* and base tag "https://www.seosazi.com/assets/" |
13
|
|
|
* and web page address "https://www.seosazi.com" |
14
|
|
|
* becomes |
15
|
|
|
* https://www.seosazi.com/assets/style.css |
16
|
|
|
* |
17
|
|
|
* Or |
18
|
|
|
*E.g. |
19
|
|
|
* href="../../style.css" |
20
|
|
|
* and base tag "https://www.seosazi.com/assets/files/" |
21
|
|
|
* and web page address "https://www.seosazi.com" |
22
|
|
|
* becomes |
23
|
|
|
* https://www.seosazi.com/style.css |
24
|
|
|
* |
25
|
|
|
* |
26
|
|
|
* Please report bugs on https://github.com/seosazi/path-converter/issues |
27
|
|
|
* |
28
|
|
|
* @author Pouya Pormohamad <[email protected]> |
29
|
|
|
* @copyright Copyright (c) 2020, Pouya Pormohamad. All rights reserved |
30
|
|
|
* @license MIT License |
31
|
|
|
*/ |
32
|
|
|
class ConvertToAbsolutePath implements ConverterInterface |
33
|
|
|
{ |
34
|
|
|
|
35
|
|
|
/** @var string */ |
36
|
|
|
private $pagePath; |
37
|
|
|
/** @var string|null */ |
38
|
|
|
private $baseTag = null; |
39
|
|
|
/** @var string|null */ |
40
|
|
|
private $starterPath=null; |
41
|
|
|
/** |
42
|
|
|
* @var array|false|int|string|null |
43
|
|
|
*/ |
44
|
|
|
private $baseTagParsing; |
45
|
|
|
/** |
46
|
|
|
* @var array|false|int|string|null |
47
|
|
|
*/ |
48
|
|
|
private $pagePathParsing; |
49
|
|
|
/** |
50
|
|
|
* @var string |
51
|
|
|
*/ |
52
|
|
|
private $domain; |
53
|
|
|
private $scheme; |
54
|
|
|
|
55
|
|
|
/** |
56
|
|
|
* ConvertToAbsolutePath constructor. |
57
|
|
|
* @param $pagePath |
58
|
|
|
*/ |
59
|
|
|
public function __construct($pagePath=NULL) |
60
|
|
|
{ |
61
|
|
|
if(isset($pagePath)) { |
62
|
|
|
$this->setPagePath($pagePath); |
63
|
|
|
} |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* @inheritDoc |
68
|
|
|
*/ |
69
|
|
|
public function convert($path) |
70
|
|
|
{ |
71
|
|
|
// Skip converting if the relative url like http://... or android-app://... etc. |
72
|
|
|
if (preg_match('/[a-z0-9-]{1,}(:\/\/)/i', $path)) { |
73
|
|
|
if(preg_match('/services:\/\//i', $path)) |
74
|
|
|
return ''; |
75
|
|
|
if(preg_match('/whatsapp:\/\//i', $path)) |
76
|
|
|
return ''; |
77
|
|
|
if(preg_match('/tel:/i', $path)) |
78
|
|
|
return ''; |
79
|
|
|
return $path; |
80
|
|
|
} |
81
|
|
|
// Treat path as invalid if it is like javascript:... etc. |
82
|
|
|
if (preg_match('/^[a-zA-Z]{0,}:[^\/]{0,1}/i', $path)) { |
83
|
|
|
return ''; |
84
|
|
|
} |
85
|
|
|
// Convert //www.google.com to http://www.google.com |
86
|
|
|
if(substr($path, 0, 2) == '//') { |
87
|
|
|
return 'http:' . $path; |
88
|
|
|
} |
89
|
|
|
// If the path is a fragment or query string, |
90
|
|
|
// it will be appended to the base url |
91
|
|
|
if($this->isHaveQueryOrFragment($path)) { |
92
|
|
|
return $this->getStarterPath() . $path; |
93
|
|
|
} |
94
|
|
|
// Treat paths with doc root, i.e, /about |
95
|
|
|
if(substr($path, 0, 1) == '/') { |
96
|
|
|
return $this->onlySitePath($this->getStarterPath()) . $path; |
97
|
|
|
} |
98
|
|
|
// For paths like ./foo, it will be appended to the furthest directory |
99
|
|
|
if(substr($path, 0, 2) == './') { |
100
|
|
|
return $this->uptoLastDir($this->getStarterPath()) . substr($path, 2); |
101
|
|
|
} |
102
|
|
|
// Convert paths like ../foo or ../../bar |
103
|
|
|
if(substr($path, 0, 3) == '../') { |
104
|
|
|
$removeTwoPointSlash = new RemovePathWithPointPointSlash($this, $path); |
105
|
|
|
return $removeTwoPointSlash->compute(); |
106
|
|
|
} |
107
|
|
|
if (empty($path)) { |
108
|
|
|
return $this->getPagePath(); |
109
|
|
|
} |
110
|
|
|
// else |
111
|
|
|
return $this->uptoLastDir($this->getStarterPath()) . $path; |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
public function onlySitePath($url) { |
115
|
|
|
// $url = preg_replace('/(^https?:\/\/.+?\/)(.*)$/i', '$1', $url); |
116
|
|
|
// return rtrim($url, '/'); |
117
|
|
|
$parseUrl = parse_url($url); |
118
|
|
|
if ($this->isCorrectUrl($parseUrl)) { |
119
|
|
|
return ''; |
120
|
|
|
}elseif(isset($parseUrl['scheme'])){ |
121
|
|
|
return $parseUrl['scheme'] . '://' . $parseUrl['host']; |
122
|
|
|
} else { |
123
|
|
|
return $parseUrl['host']; |
124
|
|
|
} |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
// Get the path with last directory |
128
|
|
|
// http://example.com/some/fake/path/page.html => http://example.com/some/fake/path/ |
129
|
|
|
public function upToLastDir($url) { |
130
|
|
|
$parseUrl = parse_url($url); |
131
|
|
|
$path = ''; |
132
|
|
|
if(isset($parseUrl['path'])) |
133
|
|
|
$path = preg_replace('/\/([^\/]+)$/i', '', $parseUrl['path']); |
134
|
|
|
return rtrim($parseUrl['scheme'] . '://' . $parseUrl['host'] . $path, '/') . '/'; |
135
|
|
|
} |
136
|
|
|
|
137
|
|
|
|
138
|
|
|
public function getPagePath() |
139
|
|
|
{ |
140
|
|
|
return $this->pagePath; |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
/** |
144
|
|
|
* @param string $pagePath |
145
|
|
|
*/ |
146
|
|
|
public function setPagePath(string $pagePath): void |
147
|
|
|
{ |
148
|
|
|
$this->pagePath = $pagePath; |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
|
152
|
|
|
public function getBaseTag() |
153
|
|
|
{ |
154
|
|
|
return $this->baseTag; |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
/** |
158
|
|
|
* @param string $baseTag |
159
|
|
|
*/ |
160
|
|
|
public function setBaseTag(string $baseTag): void |
161
|
|
|
{ |
162
|
|
|
$this->baseTag = $baseTag; |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
/** |
166
|
|
|
* @return string |
167
|
|
|
*/ |
168
|
|
|
public function getStarterPath(): string |
169
|
|
|
{ |
170
|
|
|
if($this->starterPath===null){ |
171
|
|
|
if($this->getBaseTag() === null) { |
172
|
|
|
$this->starterPath =$this->getPagePath(); |
173
|
|
|
}elseif(array_key_exists('scheme', $this->getBaseTagParsing())){ |
|
|
|
|
174
|
|
|
$this->starterPath = $this->getBaseTag() ; |
175
|
|
|
}else{ |
176
|
|
|
$this->starterPath = $this->getPagePathParsing()['scheme'] . '://' . $this->getPagePathParsing()['host'] . $this->getBaseTag(); |
177
|
|
|
} |
178
|
|
|
} |
179
|
|
|
return $this->starterPath; |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
public function getDomain() |
183
|
|
|
{ |
184
|
|
|
if ($this->domain === null) { |
185
|
|
|
if($this->getBaseTag() === null) { |
186
|
|
|
$this->domain = $this->getPagePathParsing()['host'] . '/'; |
187
|
|
|
}elseif(array_key_exists('scheme', $this->getBaseTagParsing())){ |
|
|
|
|
188
|
|
|
$this->domain = $this->getBaseTagParsing()['host'] . '/'; |
189
|
|
|
}else{ |
190
|
|
|
$this->domain = $this->getPagePathParsing()['host'] . '/'; |
191
|
|
|
} |
192
|
|
|
} |
193
|
|
|
return $this->domain; |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
public function getScheme() |
197
|
|
|
{ |
198
|
|
|
if ($this->scheme === null) { |
199
|
|
|
if($this->getBaseTag() === null) { |
200
|
|
|
$this->scheme = $this->getPagePathParsing()['scheme']; |
201
|
|
|
}elseif(array_key_exists('scheme', $this->getBaseTagParsing())){ |
|
|
|
|
202
|
|
|
$this->scheme = $this->getBaseTagParsing()['scheme']; |
203
|
|
|
}else{ |
204
|
|
|
$this->scheme = $this->getPagePathParsing()['scheme']; |
205
|
|
|
} |
206
|
|
|
} |
207
|
|
|
return $this->scheme; |
208
|
|
|
} |
209
|
|
|
|
210
|
|
|
public function getBaseTagParsing() |
211
|
|
|
{ |
212
|
|
|
if($this->baseTagParsing == null) |
213
|
|
|
$this->baseTagParsing = parse_url($this->getBaseTag()); |
214
|
|
|
return $this->baseTagParsing; |
215
|
|
|
} |
216
|
|
|
|
217
|
|
|
public function getPagePathParsing() |
218
|
|
|
{ |
219
|
|
|
if($this->pagePathParsing == null) |
220
|
|
|
$this->pagePathParsing = parse_url($this->getPagePath()); |
221
|
|
|
return $this->pagePathParsing; |
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
/** |
225
|
|
|
* @param $path |
226
|
|
|
* @return bool |
227
|
|
|
*/ |
228
|
|
|
public function isHaveQueryOrFragment($path): bool |
229
|
|
|
{ |
230
|
|
|
return substr($path, 0, 1) == '#' || substr($path, 0, 1) == '?'; |
231
|
|
|
} |
232
|
|
|
|
233
|
|
|
/** |
234
|
|
|
* @param $parseUrl |
235
|
|
|
* @return bool |
236
|
|
|
*/ |
237
|
|
|
public function isCorrectUrl($parseUrl): bool |
238
|
|
|
{ |
239
|
|
|
return !isset($parseUrl['scheme']) AND !isset($parseUrl['host']); |
240
|
|
|
} |
241
|
|
|
} |