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
|
|
|
/** |
36
|
|
|
* @var string |
37
|
|
|
*/ |
38
|
|
|
private $pagePath; |
39
|
|
|
/** |
40
|
|
|
* @var string|null |
41
|
|
|
*/ |
42
|
|
|
private $baseTag = null; |
43
|
|
|
/** |
44
|
|
|
* @var string|null |
45
|
|
|
*/ |
46
|
|
|
private $starterPath = null; |
47
|
|
|
/** |
48
|
|
|
* @var array|false|int|string|null |
49
|
|
|
*/ |
50
|
|
|
private $baseTagParsing; |
51
|
|
|
/** |
52
|
|
|
* @var array|false|int|string|null |
53
|
|
|
*/ |
54
|
|
|
private $pagePathParsing; |
55
|
|
|
/** |
56
|
|
|
* @var string |
57
|
|
|
*/ |
58
|
|
|
private $domain; |
59
|
|
|
/** |
60
|
|
|
* @var string |
61
|
|
|
*/ |
62
|
|
|
private $scheme; |
63
|
|
|
|
64
|
|
|
/** |
65
|
|
|
* ConvertToAbsolutePath constructor. |
66
|
|
|
* @param $pagePath |
67
|
|
|
* @throws \Exception |
68
|
|
|
*/ |
69
|
|
|
public function __construct($pagePath) |
70
|
|
|
{ |
71
|
|
|
if ($this->isCorrectUrl(parse_url($pagePath))) { |
72
|
|
|
$this->setPagePath($pagePath); |
73
|
|
|
}else{ |
74
|
|
|
throw new \Exception('path is not correct url'); |
75
|
|
|
} |
76
|
|
|
} |
77
|
|
|
|
78
|
|
|
/** |
79
|
|
|
* @inheritDoc |
80
|
|
|
*/ |
81
|
|
|
public function convert($path) |
82
|
|
|
{ |
83
|
|
|
// Skip converting if the relative url like http://... or android-app://... etc. |
84
|
|
|
if ($this->isPathAbsoluteOrForAnotherApp($path)) { |
85
|
|
|
return $this->checkPathIsAbsoluteOrForAnotherApp($path); |
86
|
|
|
} |
87
|
|
|
// Treat path as invalid if it is like javascript:... etc. |
88
|
|
|
if ($this->isPathJavaScript($path)) { |
89
|
|
|
return ''; |
90
|
|
|
} |
91
|
|
|
// Convert //www.google.com to http://www.google.com |
92
|
|
|
if ($this->isPathWithoutScheme($path)) { |
93
|
|
|
return 'http:' . $path; |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
// If the path is a fragment or query string, |
97
|
|
|
// it will be appended to the base url |
98
|
|
|
if ($this->isHaveQueryOrFragment($path)) { |
99
|
|
|
return $this->getStarterPath() . $path; |
100
|
|
|
} |
101
|
|
|
// Treat paths with doc root, i.e, /about |
102
|
|
|
if ($this->isPathStartWithSlash($path)) { |
103
|
|
|
return $this->onlySitePath($this->getStarterPath()) . $path; |
104
|
|
|
} |
105
|
|
|
// For paths like ./foo, it will be appended to the furthest directory |
106
|
|
|
if ($this->isPathStartWithPointSlash($path)) { |
107
|
|
|
return $this->uptoLastDir($this->getStarterPath()) . substr($path, 2); |
108
|
|
|
} |
109
|
|
|
// Convert paths like ../foo or ../../bar |
110
|
|
|
if ($this->isPathStartWithTwoPointSlash($path)) { |
111
|
|
|
return (new RemovePathWithPointPointSlash($this, $path))->compute(); |
112
|
|
|
} |
113
|
|
|
if (empty($path)) { |
114
|
|
|
return $this->getPagePath(); |
115
|
|
|
} |
116
|
|
|
// else |
117
|
|
|
return $this->uptoLastDir($this->getStarterPath()) . $path; |
118
|
|
|
} |
119
|
|
|
|
120
|
|
|
public function onlySitePath($url) { |
121
|
|
|
$parseUrl = parse_url($url); |
122
|
|
|
if ($this->isCorrectUrl($parseUrl)) { |
123
|
|
|
return $parseUrl['scheme'] . '://' . $parseUrl['host']; |
124
|
|
|
} elseif(isset($parseUrl['host'])) { |
125
|
|
|
return $parseUrl['host']; |
126
|
|
|
} else { |
127
|
|
|
return ''; |
128
|
|
|
} |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
// Get the path with last directory |
132
|
|
|
// http://example.com/some/fake/path/page.html => http://example.com/some/fake/path/ |
133
|
|
|
public function upToLastDir($url) { |
134
|
|
|
$parseUrl = parse_url($url); |
135
|
|
|
$path = ''; |
136
|
|
|
if(isset($parseUrl['path'])) { |
137
|
|
|
$path = preg_replace('/\/([^\/]+)$/i', '', $parseUrl['path']); |
138
|
|
|
} |
139
|
|
|
return rtrim($parseUrl['scheme'] . '://' . $parseUrl['host'] . $path, '/') . '/'; |
140
|
|
|
} |
141
|
|
|
|
142
|
|
|
|
143
|
|
|
public function getPagePath() |
144
|
|
|
{ |
145
|
|
|
return $this->pagePath; |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
/** |
149
|
|
|
* @param string $pagePath |
150
|
|
|
*/ |
151
|
|
|
public function setPagePath(string $pagePath): void |
152
|
|
|
{ |
153
|
|
|
$this->pagePath = $pagePath; |
154
|
|
|
} |
155
|
|
|
|
156
|
|
|
|
157
|
|
|
public function getBaseTag() |
158
|
|
|
{ |
159
|
|
|
return $this->baseTag; |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
/** |
163
|
|
|
* @param string $baseTag |
164
|
|
|
*/ |
165
|
|
|
public function setBaseTag(string $baseTag): void |
166
|
|
|
{ |
167
|
|
|
$this->baseTag = $baseTag; |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
/** |
171
|
|
|
* @return string |
172
|
|
|
*/ |
173
|
|
|
public function getStarterPath(): string |
174
|
|
|
{ |
175
|
|
|
if ($this->starterPath===null) { |
176
|
|
|
if ($this->getBaseTag() === null) { |
177
|
|
|
$this->starterPath =$this->getPagePath(); |
178
|
|
|
} elseif (array_key_exists('scheme', $this->getBaseTagParsing())) { |
|
|
|
|
179
|
|
|
$this->starterPath = $this->getBaseTag() ; |
180
|
|
|
} else { |
181
|
|
|
$this->starterPath = $this->getPagePathParsing()['scheme'] . '://' . $this->getPagePathParsing()['host'] . |
182
|
|
|
'/' . trim($this->getBaseTag(), '/') . '/'; |
183
|
|
|
} |
184
|
|
|
} |
185
|
|
|
return $this->starterPath; |
186
|
|
|
} |
187
|
|
|
|
188
|
|
|
public function getDomain() |
189
|
|
|
{ |
190
|
|
|
if ($this->domain === null) { |
191
|
|
|
if ($this->getBaseTag() === null) { |
192
|
|
|
$this->domain = $this->getPagePathParsing()['host'] . '/'; |
193
|
|
|
}elseif (array_key_exists('scheme', $this->getBaseTagParsing())){ |
|
|
|
|
194
|
|
|
$this->domain = $this->getBaseTagParsing()['host'] . '/'; |
195
|
|
|
}else { |
196
|
|
|
$this->domain = $this->getPagePathParsing()['host'] . '/'; |
197
|
|
|
} |
198
|
|
|
} |
199
|
|
|
return $this->domain; |
200
|
|
|
} |
201
|
|
|
|
202
|
|
|
public function getScheme() |
203
|
|
|
{ |
204
|
|
|
if ($this->scheme === null) { |
205
|
|
|
if ($this->getBaseTag() === null) { |
206
|
|
|
$this->scheme = $this->getPagePathParsing()['scheme']; |
207
|
|
|
}elseif (array_key_exists('scheme', $this->getBaseTagParsing())){ |
|
|
|
|
208
|
|
|
$this->scheme = $this->getBaseTagParsing()['scheme']; |
209
|
|
|
}else { |
210
|
|
|
$this->scheme = $this->getPagePathParsing()['scheme']; |
211
|
|
|
} |
212
|
|
|
} |
213
|
|
|
return $this->scheme; |
214
|
|
|
} |
215
|
|
|
|
216
|
|
|
public function getBaseTagParsing() |
217
|
|
|
{ |
218
|
|
|
if ($this->baseTagParsing == null) { |
219
|
|
|
if (is_string($this->getBaseTag())) { |
220
|
|
|
$this->baseTagParsing = parse_url($this->getBaseTag()); |
221
|
|
|
} else { |
222
|
|
|
$this->baseTagParsing = []; |
223
|
|
|
} |
224
|
|
|
} |
225
|
|
|
return $this->baseTagParsing; |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
public function getPagePathParsing() |
229
|
|
|
{ |
230
|
|
|
if ($this->pagePathParsing == null) { |
231
|
|
|
$this->pagePathParsing = parse_url($this->getPagePath()); |
232
|
|
|
} |
233
|
|
|
return $this->pagePathParsing; |
234
|
|
|
} |
235
|
|
|
|
236
|
|
|
/** |
237
|
|
|
* @param $path |
238
|
|
|
* @return bool |
239
|
|
|
*/ |
240
|
|
|
public function isHaveQueryOrFragment($path): bool |
241
|
|
|
{ |
242
|
|
|
return substr($path, 0, 1) == '#' || substr($path, 0, 1) == '?'; |
243
|
|
|
} |
244
|
|
|
|
245
|
|
|
/** |
246
|
|
|
* @param $parseUrl |
247
|
|
|
* @return bool |
248
|
|
|
*/ |
249
|
|
|
public function isCorrectUrl($parseUrl): bool |
250
|
|
|
{ |
251
|
|
|
return isset($parseUrl['scheme']) AND isset($parseUrl['host']); |
252
|
|
|
} |
253
|
|
|
|
254
|
|
|
/** |
255
|
|
|
* @param $path |
256
|
|
|
* @return string |
257
|
|
|
*/ |
258
|
|
|
public function checkPathIsAbsoluteOrForAnotherApp($path): string |
259
|
|
|
{ |
260
|
|
|
if (preg_match('/services:\/\//i', $path)) { |
261
|
|
|
return ''; |
262
|
|
|
} |
263
|
|
|
if (preg_match('/whatsapp:\/\//i', $path)) { |
264
|
|
|
return ''; |
265
|
|
|
} |
266
|
|
|
if (preg_match('/tel:/i', $path)) { |
267
|
|
|
return ''; |
268
|
|
|
} |
269
|
|
|
return $path; |
270
|
|
|
} |
271
|
|
|
|
272
|
|
|
/** |
273
|
|
|
* @param $path |
274
|
|
|
* @return false|int |
275
|
|
|
*/ |
276
|
|
|
public function isPathAbsoluteOrForAnotherApp($path) |
277
|
|
|
{ |
278
|
|
|
return preg_match('/[a-z0-9-]{1,}(:\/\/)/i', $path); |
279
|
|
|
} |
280
|
|
|
|
281
|
|
|
/** |
282
|
|
|
* @param $path |
283
|
|
|
* @return false|int |
284
|
|
|
*/ |
285
|
|
|
public function isPathJavaScript($path) |
286
|
|
|
{ |
287
|
|
|
return preg_match('/^[a-zA-Z]{0,}:[^\/]{0,1}/i', $path); |
288
|
|
|
} |
289
|
|
|
|
290
|
|
|
/** |
291
|
|
|
* @param $path |
292
|
|
|
* @return bool |
293
|
|
|
*/ |
294
|
|
|
public function isPathWithoutScheme($path): bool |
295
|
|
|
{ |
296
|
|
|
return substr($path, 0, 2) == '//'; |
297
|
|
|
} |
298
|
|
|
|
299
|
|
|
/** |
300
|
|
|
* @param $path |
301
|
|
|
* @return bool |
302
|
|
|
*/ |
303
|
|
|
public function isPathStartWithSlash($path): bool |
304
|
|
|
{ |
305
|
|
|
return substr($path, 0, 1) == '/'; |
306
|
|
|
} |
307
|
|
|
|
308
|
|
|
/** |
309
|
|
|
* @param $path |
310
|
|
|
* @return bool |
311
|
|
|
*/ |
312
|
|
|
public function isPathStartWithPointSlash($path): bool |
313
|
|
|
{ |
314
|
|
|
return substr($path, 0, 2) == './'; |
315
|
|
|
} |
316
|
|
|
|
317
|
|
|
/** |
318
|
|
|
* @param $path |
319
|
|
|
* @return bool |
320
|
|
|
*/ |
321
|
|
|
public function isPathStartWithTwoPointSlash($path): bool |
322
|
|
|
{ |
323
|
|
|
return substr($path, 0, 3) == '../'; |
324
|
|
|
} |
325
|
|
|
} |