|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* @author Gerard van Helden <[email protected]> |
|
4
|
|
|
* @copyright Zicht Online <http://zicht.nl> |
|
5
|
|
|
*/ |
|
6
|
|
|
|
|
7
|
|
|
namespace Zicht\Bundle\UrlBundle\Url; |
|
8
|
|
|
|
|
9
|
|
|
use Zicht\Bundle\UrlBundle\Aliasing\Aliasing; |
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* Class Rewriter |
|
13
|
|
|
*/ |
|
14
|
|
|
class Rewriter |
|
15
|
|
|
{ |
|
16
|
|
|
private $localDomains = []; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* Constructor |
|
20
|
|
|
* |
|
21
|
|
|
* @param Aliasing $aliasing |
|
22
|
|
|
*/ |
|
23
|
|
|
public function __construct(Aliasing $aliasing) |
|
24
|
|
|
{ |
|
25
|
|
|
$this->aliasing = $aliasing; |
|
|
|
|
|
|
26
|
|
|
} |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Set a list of domains in urls to consider 'local', i.e. process the 'path' while keeping the domain in tact. |
|
30
|
|
|
* |
|
31
|
|
|
* @param string[] $localDomains |
|
32
|
|
|
*/ |
|
33
|
|
|
public function setLocalDomains($localDomains) |
|
34
|
|
|
{ |
|
35
|
|
|
$this->localDomains = $localDomains; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Rewrite all $urls, given the $mappings map of mappings (from => to), within the context of |
|
40
|
|
|
* host names $localDomains. |
|
41
|
|
|
* |
|
42
|
|
|
* Example: |
|
43
|
|
|
* |
|
44
|
|
|
* rewrite(['http://example.org/foo?x=y'], ['/foo' => '/bar'], ['example.org']); |
|
45
|
|
|
* |
|
46
|
|
|
* |
|
47
|
|
|
* Would return the following mapping: |
|
48
|
|
|
* |
|
49
|
|
|
* ['http://example.org/foo?x=y' => 'http://example.org/bar?x=y'] |
|
50
|
|
|
* |
|
51
|
|
|
* @param string[] $urls |
|
52
|
|
|
* @param string $mode |
|
53
|
|
|
* @return array |
|
54
|
|
|
*/ |
|
55
|
|
|
public function rewrite(array $urls, $mode) |
|
56
|
|
|
{ |
|
57
|
|
|
$mappings = $this->aliasing->getAliasingMap( |
|
58
|
|
|
array_map( |
|
59
|
|
|
[$this, 'extractPath'], |
|
60
|
|
|
$urls |
|
61
|
|
|
), |
|
62
|
|
|
$mode |
|
63
|
|
|
); |
|
64
|
|
|
|
|
65
|
|
|
$ret = []; |
|
66
|
|
|
foreach ($urls as $url) { |
|
67
|
|
|
if (isset($mappings[$url])) { |
|
68
|
|
|
// early match, if the value can be mapping directly, used that. |
|
69
|
|
|
$ret[$url] = $mappings[$url]; |
|
70
|
|
|
continue; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
$parts = $this->parseUrl($url); |
|
74
|
|
|
|
|
75
|
|
|
if (!isset($parts['path'])) { |
|
76
|
|
|
// no path, nothing to map. |
|
77
|
|
|
$ret[$url] = $url; |
|
78
|
|
|
continue; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
if (isset($parts['host'])) { |
|
82
|
|
|
if (!in_array($parts['host'], $this->localDomains)) { |
|
83
|
|
|
// external url, don't do mapping. |
|
84
|
|
|
$ret[$url] = $url; |
|
85
|
|
|
continue; |
|
86
|
|
|
} |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
// don't rewrite this. |
|
90
|
|
|
if (isset($parts['user']) || isset($parts['password'])) { |
|
91
|
|
|
$ret[$url]= $url; |
|
92
|
|
|
continue; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
$rewritten = ''; |
|
96
|
|
|
if (isset($parts['scheme'])) { |
|
97
|
|
|
$rewritten .= $parts['scheme'] . ':'; |
|
98
|
|
|
} |
|
99
|
|
|
if (isset($parts['host'])) { |
|
100
|
|
|
$rewritten .= '//' . $parts['host']; |
|
101
|
|
|
} |
|
102
|
|
|
if (isset($parts['port'])) { |
|
103
|
|
|
$rewritten .= ':' . $parts['port']; |
|
104
|
|
|
} |
|
105
|
|
|
if (isset($parts['path'])) { |
|
106
|
|
|
if (isset($mappings[$parts['path']])) { |
|
107
|
|
|
$rewritten .= $mappings[$parts['path']]; |
|
108
|
|
|
} else { |
|
109
|
|
|
// no match on path level, keep as-is |
|
110
|
|
|
$ret[$url] = $url; |
|
111
|
|
|
continue; |
|
112
|
|
|
} |
|
113
|
|
|
} |
|
114
|
|
|
if (isset($parts['params'])) { |
|
115
|
|
|
$rewritten .= '/' . $parts['params']; |
|
116
|
|
|
} |
|
117
|
|
|
if (isset($parts['query'])) { |
|
118
|
|
|
$rewritten .= '?' . $parts['query']; |
|
119
|
|
|
} |
|
120
|
|
|
if (isset($parts['fragment'])) { |
|
121
|
|
|
$rewritten .= '#' . $parts['fragment']; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
$ret[$url] = $rewritten; |
|
125
|
|
|
} |
|
126
|
|
|
return $ret; |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
|
|
130
|
|
|
/** |
|
131
|
|
|
* Extract the path of the URL which is considered for aliasing. |
|
132
|
|
|
* |
|
133
|
|
|
* @param string $url |
|
134
|
|
|
* @return string|null |
|
135
|
|
|
*/ |
|
136
|
|
|
public function extractPath($url) |
|
137
|
|
|
{ |
|
138
|
|
|
$parts = $this->parseUrl($url); |
|
139
|
|
|
|
|
140
|
|
|
if (!isset($parts['path'])) { |
|
141
|
|
|
return null; |
|
142
|
|
|
} |
|
143
|
|
|
// ignore non-http schemes |
|
144
|
|
|
if (isset($parts['scheme']) && !in_array($parts['scheme'], ['http', 'https'])) { |
|
145
|
|
|
return null; |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
|
|
return $parts['path']; |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
|
|
152
|
|
|
/** |
|
153
|
|
|
* Parse the url, and additionally add a 'parameters' part which is defined as follows: |
|
154
|
|
|
* |
|
155
|
|
|
* The parameters may occur after the path, separated by slashes, where each of the key value pairs are |
|
156
|
|
|
* separated by '='. |
|
157
|
|
|
* |
|
158
|
|
|
* e.g.: |
|
159
|
|
|
* /this/is/the/path/while=these/are=paremeters |
|
160
|
|
|
* ^----- path ----^ ^------- params ---------^ |
|
161
|
|
|
* |
|
162
|
|
|
* @param string $url |
|
163
|
|
|
* @return mixed |
|
164
|
|
|
*/ |
|
165
|
|
|
public function parseUrl($url) |
|
166
|
|
|
{ |
|
167
|
|
|
$ret = parse_url($url); |
|
168
|
|
|
|
|
169
|
|
|
if (isset($ret['path'])) { |
|
170
|
|
|
$parts = explode('/', $ret['path']); |
|
171
|
|
|
$params = []; |
|
172
|
|
|
while (false !== strpos(end($parts), '=')) { |
|
173
|
|
|
array_push($params, array_pop($parts)); |
|
174
|
|
|
} |
|
175
|
|
|
if (count($params)) { |
|
176
|
|
|
$ret['path'] = join('/', $parts); |
|
177
|
|
|
$ret['params'] = join('/', $params); |
|
178
|
|
|
} |
|
179
|
|
|
} |
|
180
|
|
|
|
|
181
|
|
|
return $ret; |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
|
|
185
|
|
|
/** |
|
186
|
|
|
* This convenenience method processes the urls in the content with a structure matching the following: |
|
187
|
|
|
* |
|
188
|
|
|
* ['the url' => ['total string that should be replaced', 'the prefix', 'the url', 'the suffix']] |
|
189
|
|
|
* |
|
190
|
|
|
* e.g.: |
|
191
|
|
|
* 'http://example.org/' => ['<a href="http://example.org/"', '<a href="', 'http://example.org/', '"'] |
|
192
|
|
|
* |
|
193
|
|
|
* It replaces contents by replacing all occurrences of the 0 index with the 1 index, concatenated with the |
|
194
|
|
|
* rewritten url, and the suffix. |
|
195
|
|
|
* |
|
196
|
|
|
* @param string $content |
|
197
|
|
|
* @param string $mode |
|
198
|
|
|
* @param mixed $matchedGroups |
|
199
|
|
|
* @return string |
|
200
|
|
|
*/ |
|
201
|
|
|
public function rewriteMatches($content, $mode, $matchedGroups) |
|
202
|
|
|
{ |
|
203
|
|
|
$replacements = []; |
|
204
|
|
|
foreach ($this->rewrite(array_keys($matchedGroups), $mode) as $from => $to) { |
|
205
|
|
|
if (isset($matchedGroups[$from]) && $from !== $to) { |
|
206
|
|
|
foreach ($matchedGroups[$from] as list($source, $prefix, $oldUrl, $suffix)) { |
|
207
|
|
|
$replacements[$source] = $prefix . $to . $suffix; |
|
208
|
|
|
} |
|
209
|
|
|
} |
|
210
|
|
|
} |
|
211
|
|
|
|
|
212
|
|
|
return strtr($content, $replacements); |
|
213
|
|
|
} |
|
214
|
|
|
} |
|
215
|
|
|
|