1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* Retour plugin for Craft CMS |
4
|
|
|
* |
5
|
|
|
* Retour allows you to intelligently redirect legacy URLs, so that you don't |
6
|
|
|
* lose SEO value when rebuilding & restructuring a website |
7
|
|
|
* |
8
|
|
|
* @link https://nystudio107.com/ |
|
|
|
|
9
|
|
|
* @copyright Copyright (c) 2020 nystudio107 |
|
|
|
|
10
|
|
|
*/ |
|
|
|
|
11
|
|
|
|
12
|
|
|
namespace nystudio107\retour\helpers; |
13
|
|
|
|
14
|
|
|
use Craft; |
15
|
|
|
use craft\errors\SiteNotFoundException; |
16
|
|
|
use craft\helpers\UrlHelper as CraftUrlHelper; |
17
|
|
|
|
18
|
|
|
/** |
|
|
|
|
19
|
|
|
* @author nystudio107 |
|
|
|
|
20
|
|
|
* @package Retour |
|
|
|
|
21
|
|
|
* @since 3.1.34 |
|
|
|
|
22
|
|
|
*/ |
|
|
|
|
23
|
|
|
class UrlHelper extends CraftUrlHelper |
24
|
|
|
{ |
25
|
|
|
// Public Static Methods |
26
|
|
|
// ========================================================================= |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* Returns a query string that is a combination of al of the query strings from |
30
|
|
|
* the passed in $urls |
31
|
|
|
* |
32
|
|
|
* @param ...$urls |
|
|
|
|
33
|
|
|
* @return string |
|
|
|
|
34
|
|
|
*/ |
35
|
|
|
public static function combineQueryStringsFromUrls(...$urls): string |
36
|
|
|
{ |
37
|
|
|
$queryParams = []; |
38
|
|
|
foreach ($urls as $url) { |
39
|
|
|
$parsedUrl = parse_url($url); |
40
|
|
|
$params = []; |
41
|
|
|
parse_str($parsedUrl['query'] ?? '', $params); |
42
|
|
|
$queryParams[] = $params; |
43
|
|
|
} |
44
|
|
|
$queryParams = array_unique(array_merge([], ...$queryParams), SORT_REGULAR); |
45
|
|
|
|
46
|
|
|
return http_build_query($queryParams); |
47
|
|
|
} |
48
|
|
|
|
49
|
|
|
/** |
50
|
|
|
* Strip out any site-defined sub-path from the incoming $path |
51
|
|
|
* e.g. if the Site's baseUrl is set to https://example.com/es and $path is /es/blog |
52
|
|
|
* this function will return /blog |
53
|
|
|
* |
54
|
|
|
* @param string $path |
|
|
|
|
55
|
|
|
* @return string |
|
|
|
|
56
|
|
|
*/ |
57
|
|
|
public static function stripSitePathPrefix($path): string |
58
|
|
|
{ |
59
|
|
|
try { |
60
|
|
|
$baseSiteUrl = self::baseSiteUrl(); |
61
|
|
|
} catch (SiteNotFoundException $e) { |
62
|
|
|
$baseSiteUrl = ''; |
63
|
|
|
} |
64
|
|
|
$sitePath = parse_url($baseSiteUrl, PHP_URL_PATH); |
65
|
|
|
$addSlash = str_ends_with($path, '/'); |
66
|
|
|
if (!empty($sitePath)) { |
67
|
|
|
// Normalizes a URI path by trimming leading/ trailing slashes and removing double slashes |
68
|
|
|
$sitePath = '/' . preg_replace('/\/\/+/', '/', trim($sitePath, '/')); |
69
|
|
|
} |
70
|
|
|
// Strip the $sitePath from the incoming $path |
71
|
|
|
if (str_starts_with($path, $sitePath)) { |
72
|
|
|
$path = substr($path, strlen($sitePath)); |
73
|
|
|
$path = '/' . preg_replace('/\/\/+/', '/', trim($path, '/')); |
74
|
|
|
} |
75
|
|
|
$path = $addSlash ? $path . '/' : $path; |
76
|
|
|
|
77
|
|
|
return $path; |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
/** |
81
|
|
|
* See if the path includes a site prefix for any site |
82
|
|
|
* |
83
|
|
|
* @param string $path |
|
|
|
|
84
|
|
|
* @return bool |
|
|
|
|
85
|
|
|
*/ |
86
|
|
|
public static function pathHasSitePrefix(string $path): bool |
87
|
|
|
{ |
88
|
|
|
$sites = Craft::$app->getSites()->getAllSites(); |
89
|
|
|
foreach ($sites as $site) { |
90
|
|
|
$sitePath = parse_url($site->baseUrl, PHP_URL_PATH); |
91
|
|
|
if (!empty($sitePath)) { |
92
|
|
|
// Normalizes a URI path by trimming leading/ trailing slashes and removing double slashes |
93
|
|
|
$sitePath = '/' . preg_replace('/\/\/+/', '/', trim($sitePath, '/')); |
94
|
|
|
} |
95
|
|
|
// Strip the $sitePath from the incoming $path |
96
|
|
|
if (str_starts_with($path, $sitePath)) { |
97
|
|
|
return true; |
98
|
|
|
} |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
return false; |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
/** |
105
|
|
|
* Merge the $url and $path together, combining any overlapping path segments |
106
|
|
|
* |
107
|
|
|
* @param ?string $url |
|
|
|
|
108
|
|
|
* @param ?string $path |
|
|
|
|
109
|
|
|
* @return string |
|
|
|
|
110
|
|
|
*/ |
111
|
|
|
public static function mergeUrlWithPath($url, $path): string |
112
|
|
|
{ |
113
|
|
|
$overlap = 0; |
114
|
|
|
$url = $url ?? ''; |
115
|
|
|
$path = $path ?? ''; |
116
|
|
|
$url = rtrim($url, '/') . '/'; |
117
|
|
|
$path = '/' . ltrim($path, '/'); |
118
|
|
|
$urlOffset = strlen($url); |
119
|
|
|
$pathLength = strlen($path); |
120
|
|
|
$pathOffset = 0; |
121
|
|
|
while ($urlOffset > 0 && $pathOffset < $pathLength) { |
122
|
|
|
$urlOffset--; |
123
|
|
|
$pathOffset++; |
124
|
|
|
if (str_starts_with($path, substr($url, $urlOffset, $pathOffset))) { |
125
|
|
|
$overlap = $pathOffset; |
126
|
|
|
} |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
return rtrim($url, '/') . '/' . ltrim(substr($path, $overlap), '/'); |
130
|
|
|
} |
131
|
|
|
|
132
|
|
|
/** |
133
|
|
|
* Return a sanitized URL |
134
|
|
|
* |
135
|
|
|
* @param string $url |
|
|
|
|
136
|
|
|
* |
137
|
|
|
* @return string |
138
|
|
|
*/ |
139
|
1 |
|
public static function sanitizeUrl(string $url): string |
140
|
|
|
{ |
141
|
|
|
// HTML decode the entities, then strip out any tags |
142
|
1 |
|
$url = html_entity_decode($url, ENT_NOQUOTES, 'UTF-8'); |
143
|
1 |
|
$url = urldecode($url); |
144
|
1 |
|
$url = strip_tags($url); |
145
|
|
|
// Remove any Twig tags that somehow are present in the incoming URL |
146
|
|
|
/** @noinspection CallableParameterUseCaseInTypeContextInspection */ |
|
|
|
|
147
|
1 |
|
$url = preg_replace('/{.*}/', '', $url); |
148
|
|
|
// Remove any linebreaks that may be errantly in the URL |
149
|
1 |
|
$url = (string)str_replace([ |
|
|
|
|
150
|
1 |
|
PHP_EOL, |
151
|
1 |
|
"\r", |
152
|
1 |
|
"\n", |
153
|
1 |
|
] |
154
|
1 |
|
, '', $url |
|
|
|
|
155
|
1 |
|
); |
156
|
|
|
|
157
|
1 |
|
return $url; |
158
|
|
|
} |
159
|
|
|
} |
160
|
|
|
|