1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* SEOmatic plugin for Craft CMS 3.x |
4
|
|
|
* |
5
|
|
|
* A turnkey SEO implementation for Craft CMS that is comprehensive, powerful, |
6
|
|
|
* and flexible |
7
|
|
|
* |
8
|
|
|
* @link https://nystudio107.com |
9
|
|
|
* @copyright Copyright (c) 2017 nystudio107 |
10
|
|
|
*/ |
11
|
|
|
|
12
|
|
|
namespace nystudio107\seomatic\helpers; |
13
|
|
|
|
14
|
|
|
use Craft; |
15
|
|
|
use craft\errors\SiteNotFoundException; |
16
|
|
|
use craft\helpers\UrlHelper as CraftUrlHelper; |
17
|
|
|
use nystudio107\seomatic\Seomatic; |
18
|
|
|
use Throwable; |
19
|
|
|
use yii\base\Exception; |
20
|
|
|
use function is_string; |
21
|
|
|
|
22
|
|
|
/** |
23
|
|
|
* @author nystudio107 |
24
|
|
|
* @package Seomatic |
25
|
|
|
* @since 3.0.0 |
26
|
|
|
*/ |
27
|
|
|
class UrlHelper extends CraftUrlHelper |
28
|
|
|
{ |
29
|
|
|
// Public Static Properties |
30
|
|
|
// ========================================================================= |
31
|
|
|
|
32
|
|
|
// Public Static Methods |
33
|
|
|
// ========================================================================= |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* @inheritDoc |
37
|
|
|
*/ |
38
|
|
|
public static function siteUrl(string $path = '', $params = null, string $scheme = null, int $siteId = null): string |
39
|
|
|
{ |
40
|
|
|
try { |
41
|
|
|
$siteUrl = self::getSiteUrlOverrideSetting($siteId); |
42
|
|
|
} catch (Throwable $e) { |
43
|
|
|
// That's okay |
44
|
|
|
} |
45
|
|
|
if (!empty($siteUrl)) { |
46
|
|
|
$siteUrl = MetaValue::parseString($siteUrl); |
47
|
|
|
// Extract out just the path part |
48
|
|
|
$parts = self::decomposeUrl($path); |
49
|
|
|
$path = $parts['path'] . $parts['suffix']; |
50
|
|
|
$url = self::mergeUrlWithPath($siteUrl, $path); |
51
|
|
|
// Handle trailing slashes properly for generated URLs |
52
|
|
|
$generalConfig = Craft::$app->getConfig()->getGeneral(); |
53
|
|
|
if ($generalConfig->addTrailingSlashesToUrls && !preg_match('/(.+\?.*)|(\.[^\/]+$)/', $url)) { |
54
|
|
|
$url = rtrim($url, '/') . '/'; |
55
|
|
|
} |
56
|
|
|
if (!$generalConfig->addTrailingSlashesToUrls) { |
57
|
|
|
$url = rtrim($url, '/'); |
58
|
|
|
} |
59
|
|
|
|
60
|
|
|
return DynamicMeta::sanitizeUrl(parent::urlWithParams($url, $params ?? []), false, false); |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
return DynamicMeta::sanitizeUrl(parent::siteUrl($path, $params, $scheme, $siteId), false, false); |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* Merge the $url and $path together, combining any overlapping path segments |
68
|
|
|
* |
69
|
|
|
* @param string $url |
70
|
|
|
* @param string $path |
71
|
|
|
* @return string |
72
|
|
|
*/ |
73
|
|
|
public static function mergeUrlWithPath(string $url, string $path): string |
74
|
|
|
{ |
75
|
|
|
$overlap = 0; |
76
|
|
|
$url = rtrim($url, '/') . '/'; |
77
|
|
|
$path = '/' . ltrim($path, '/'); |
78
|
|
|
$urlOffset = strlen($url); |
79
|
|
|
$pathLength = strlen($path); |
80
|
|
|
$pathOffset = 0; |
81
|
|
|
while ($urlOffset > 0 && $pathOffset < $pathLength) { |
82
|
|
|
$urlOffset--; |
83
|
|
|
$pathOffset++; |
84
|
|
|
if (str_starts_with($path, substr($url, $urlOffset, $pathOffset))) { |
85
|
|
|
$overlap = $pathOffset; |
86
|
|
|
} |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
return rtrim($url, '/') . '/' . ltrim(substr($path, $overlap), '/'); |
90
|
|
|
} |
91
|
|
|
|
92
|
|
|
/** |
93
|
|
|
* Return the page trigger and the value of the page trigger (null if it doesn't exist) |
94
|
|
|
* |
95
|
|
|
* @return array |
96
|
|
|
*/ |
97
|
|
|
public static function pageTriggerValue(): array |
98
|
|
|
{ |
99
|
|
|
$pageTrigger = Craft::$app->getConfig()->getGeneral()->pageTrigger; |
100
|
|
|
if (!is_string($pageTrigger) || $pageTrigger === '') { |
101
|
|
|
$pageTrigger = 'p'; |
102
|
|
|
} |
103
|
|
|
// Is this query string-based pagination? |
104
|
|
|
if ($pageTrigger[0] === '?') { |
105
|
|
|
$pageTrigger = trim($pageTrigger, '?='); |
106
|
|
|
} |
107
|
|
|
// Avoid conflict with the path param |
108
|
|
|
$pathParam = Craft::$app->getConfig()->getGeneral()->pathParam; |
109
|
|
|
if ($pageTrigger === $pathParam) { |
110
|
|
|
$pageTrigger = $pathParam === 'p' ? 'pg' : 'p'; |
111
|
|
|
} |
112
|
|
|
$pageTriggerValue = Craft::$app->getRequest()->getParam($pageTrigger); |
113
|
|
|
|
114
|
|
|
return [$pageTrigger, $pageTriggerValue]; |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
/** |
118
|
|
|
* Return an absolute URL with protocol that curl will be happy with |
119
|
|
|
* |
120
|
|
|
* @param string $url |
121
|
|
|
* |
122
|
|
|
* @return string |
123
|
|
|
*/ |
124
|
|
|
public static function absoluteUrlWithProtocol($url): string |
125
|
|
|
{ |
126
|
|
|
// Make this a full URL |
127
|
|
|
if (!self::isAbsoluteUrl($url)) { |
128
|
|
|
$protocol = 'http'; |
129
|
|
|
if (isset($_SERVER['HTTPS']) && (strcasecmp($_SERVER['HTTPS'], 'on') === 0 || $_SERVER['HTTPS'] == 1) |
|
|
|
|
130
|
|
|
|| isset($_SERVER['HTTP_X_FORWARDED_PROTO']) && strcasecmp($_SERVER['HTTP_X_FORWARDED_PROTO'], 'https') === 0 |
131
|
|
|
) { |
132
|
|
|
$protocol = 'https'; |
133
|
|
|
} |
134
|
|
|
if (self::isProtocolRelativeUrl($url)) { |
135
|
|
|
try { |
136
|
|
|
$url = self::urlWithScheme($url, $protocol); |
137
|
|
|
} catch (SiteNotFoundException $e) { |
138
|
|
|
Craft::error($e->getMessage(), __METHOD__); |
139
|
|
|
} |
140
|
|
|
} else { |
141
|
|
|
try { |
142
|
|
|
$url = self::siteUrl($url, null, $protocol); |
143
|
|
|
if (self::isProtocolRelativeUrl($url)) { |
144
|
|
|
$url = self::urlWithScheme($url, $protocol); |
145
|
|
|
} |
146
|
|
|
} catch (Exception $e) { |
147
|
|
|
Craft::error($e->getMessage(), __METHOD__); |
148
|
|
|
} |
149
|
|
|
} |
150
|
|
|
} |
151
|
|
|
// Ensure that any spaces in the URL are encoded |
152
|
|
|
$url = str_replace(' ', '%20', $url); |
153
|
|
|
// If the incoming URL has a trailing slash, respect it by preserving it |
154
|
|
|
$preserveTrailingSlash = false; |
155
|
|
|
if (str_ends_with($url, '/')) { |
156
|
|
|
$preserveTrailingSlash = true; |
157
|
|
|
} |
158
|
|
|
// Handle trailing slashes properly for generated URLs |
159
|
|
|
$generalConfig = Craft::$app->getConfig()->getGeneral(); |
160
|
|
|
if ($generalConfig->addTrailingSlashesToUrls && !preg_match('/(.+\?.*)|(\.[^\/]+$)/', $url)) { |
161
|
|
|
$url = rtrim($url, '/') . '/'; |
162
|
|
|
} |
163
|
|
|
if (!$generalConfig->addTrailingSlashesToUrls && (!$preserveTrailingSlash || self::urlIsSiteIndex($url))) { |
164
|
|
|
$url = rtrim($url, '/'); |
165
|
|
|
} |
166
|
|
|
|
167
|
|
|
return DynamicMeta::sanitizeUrl($url, false, false); |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
/** |
171
|
|
|
* urlencode() just the query parameters in the URL |
172
|
|
|
* |
173
|
|
|
* @param string $url |
174
|
|
|
* @return string |
175
|
|
|
*/ |
176
|
2 |
|
public static function encodeUrlQueryParams(string $url): string |
177
|
|
|
{ |
178
|
2 |
|
$urlParts = parse_url($url); |
179
|
2 |
|
$encodedUrl = ""; |
180
|
2 |
|
if (isset($urlParts['scheme'])) { |
181
|
|
|
$encodedUrl .= $urlParts['scheme'] . '://'; |
182
|
|
|
} |
183
|
2 |
|
if (isset($urlParts['host'])) { |
184
|
|
|
$encodedUrl .= $urlParts['host']; |
185
|
|
|
} |
186
|
2 |
|
if (isset($urlParts['port'])) { |
187
|
|
|
$encodedUrl .= ':' . $urlParts['port']; |
188
|
|
|
} |
189
|
2 |
|
if (isset($urlParts['path'])) { |
190
|
2 |
|
$encodedUrl .= $urlParts['path']; |
191
|
|
|
} |
192
|
2 |
|
if (isset($urlParts['query'])) { |
193
|
|
|
$query = explode('&', $urlParts['query']); |
194
|
|
|
foreach ($query as $j => $value) { |
195
|
|
|
$value = explode('=', $value, 2); |
196
|
|
|
if (count($value) === 2) { |
197
|
|
|
$query[$j] = urlencode($value[0]) . '=' . urlencode($value[1]); |
198
|
|
|
} else { |
199
|
|
|
$query[$j] = urlencode($value[0]); |
200
|
|
|
} |
201
|
|
|
} |
202
|
|
|
$encodedUrl .= '?' . implode('&', $query); |
203
|
|
|
} |
204
|
2 |
|
if (isset($urlParts['fragment'])) { |
205
|
|
|
$encodedUrl .= '#' . $urlParts['fragment']; |
206
|
|
|
} |
207
|
|
|
|
208
|
2 |
|
return $encodedUrl; |
209
|
|
|
} |
210
|
|
|
|
211
|
|
|
/** |
212
|
|
|
* Return whether this URL has a sub-directory as part of it |
213
|
|
|
* |
214
|
|
|
* @param string $url |
215
|
|
|
* @return bool |
216
|
|
|
*/ |
217
|
|
|
public static function urlHasSubDir(string $url): bool |
218
|
|
|
{ |
219
|
|
|
return !empty(parse_url(trim($url, '/'), PHP_URL_PATH)); |
220
|
|
|
} |
221
|
|
|
|
222
|
|
|
/** |
223
|
|
|
* See if the url is a site index, and if so, strip the trailing slash |
224
|
|
|
* ref: https://github.com/craftcms/cms/issues/5675 |
225
|
|
|
* |
226
|
|
|
* @param string $url |
227
|
|
|
* @return bool |
228
|
|
|
*/ |
229
|
|
|
public static function urlIsSiteIndex(string $url): bool |
230
|
|
|
{ |
231
|
|
|
$sites = Craft::$app->getSites()->getAllSites(); |
232
|
|
|
$result = false; |
233
|
|
|
foreach ($sites as $site) { |
234
|
|
|
$sitePath = parse_url(self::siteUrl('/', null, null, $site->id), PHP_URL_PATH); |
235
|
|
|
if (!empty($sitePath)) { |
236
|
|
|
// Normalizes a URI path by trimming leading/ trailing slashes and removing double slashes |
237
|
|
|
$sitePath = '/' . preg_replace('/\/\/+/', '/', trim($sitePath, '/')); |
238
|
|
|
} |
239
|
|
|
// Normalizes a URI path by trimming leading/ trailing slashes and removing double slashes |
240
|
|
|
$url = '/' . preg_replace('/\/\/+/', '/', trim($url, '/')); |
241
|
|
|
// See if this url ends with a site prefix, and thus is a site index |
242
|
|
|
if (str_ends_with($url, $sitePath)) { |
243
|
|
|
$result = true; |
244
|
|
|
} |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
return $result; |
248
|
|
|
} |
249
|
|
|
|
250
|
|
|
/** |
251
|
|
|
* Return the siteUrlOverride setting, which can be a string or an array of site URLs |
252
|
|
|
* indexed by the site handle |
253
|
|
|
* |
254
|
|
|
* @param int|null $siteId |
255
|
|
|
* @return string |
256
|
|
|
* @throws Exception |
257
|
|
|
* @throws SiteNotFoundException |
258
|
|
|
*/ |
259
|
|
|
public static function getSiteUrlOverrideSetting(?int $siteId = null): string |
260
|
|
|
{ |
261
|
|
|
// If the override is a string, just return it |
262
|
|
|
$siteUrlOverride = Seomatic::$settings->siteUrlOverride; |
263
|
|
|
if (is_string($siteUrlOverride)) { |
264
|
|
|
return $siteUrlOverride; |
265
|
|
|
} |
266
|
|
|
// If the override is an array, pluck the appropriate one by handle |
267
|
|
|
if (is_array($siteUrlOverride)) { |
|
|
|
|
268
|
|
|
$sites = Craft::$app->getSites(); |
269
|
|
|
$site = $sites->getCurrentSite(); |
270
|
|
|
if ($siteId !== null) { |
271
|
|
|
$site = $sites->getSiteById($siteId, true); |
272
|
|
|
if (!$site) { |
273
|
|
|
throw new Exception('Invalid site ID: ' . $siteId); |
274
|
|
|
} |
275
|
|
|
} |
276
|
|
|
|
277
|
|
|
return $siteUrlOverride[$site->handle] ?? ''; |
278
|
|
|
} |
|
|
|
|
279
|
|
|
} |
280
|
|
|
|
281
|
|
|
/** |
282
|
|
|
* Encodes non-alphanumeric characters in a URL, except reserved characters and already-encoded characters. |
283
|
|
|
* |
284
|
|
|
* @param string $url |
285
|
|
|
* @return string |
286
|
|
|
* @since 4.13.0 |
287
|
|
|
*/ |
288
|
|
|
public static function encodeUrl(string $url): string |
289
|
|
|
{ |
290
|
|
|
$parts = preg_split('/([:\/?#\[\]@!$&\'()*+,;=%])/', $url, -1, PREG_SPLIT_DELIM_CAPTURE); |
291
|
|
|
$url = ''; |
292
|
|
|
foreach ($parts as $i => $part) { |
293
|
|
|
if ($i % 2 === 0) { |
294
|
|
|
$url .= urlencode($part); |
295
|
|
|
} else { |
296
|
|
|
$url .= $part; |
297
|
|
|
} |
298
|
|
|
} |
299
|
|
|
return $url; |
300
|
|
|
} |
301
|
|
|
|
302
|
|
|
// Protected Methods |
303
|
|
|
// ========================================================================= |
304
|
|
|
|
305
|
|
|
/** |
306
|
|
|
* Decompose a url into a prefix, path, and suffix |
307
|
|
|
* |
308
|
|
|
* @param $pathOrUrl |
309
|
|
|
* |
310
|
|
|
* @return array |
311
|
|
|
*/ |
312
|
|
|
protected static function decomposeUrl($pathOrUrl): array |
313
|
|
|
{ |
314
|
|
|
$result = array(); |
315
|
|
|
|
316
|
|
|
if (filter_var($pathOrUrl, FILTER_VALIDATE_URL)) { |
317
|
|
|
$url_parts = parse_url($pathOrUrl); |
318
|
|
|
$result['prefix'] = $url_parts['scheme'] . '://' . $url_parts['host']; |
319
|
|
|
$result['path'] = $url_parts['path'] ?? ''; |
320
|
|
|
$result['suffix'] = ''; |
321
|
|
|
$result['suffix'] .= empty($url_parts['query']) ? '' : '?' . $url_parts['query']; |
322
|
|
|
$result['suffix'] .= empty($url_parts['fragment']) ? '' : '#' . $url_parts['fragment']; |
323
|
|
|
} else { |
324
|
|
|
$result['prefix'] = ''; |
325
|
|
|
$result['path'] = $pathOrUrl; |
326
|
|
|
$result['suffix'] = ''; |
327
|
|
|
} |
328
|
|
|
|
329
|
|
|
return $result; |
330
|
|
|
} |
331
|
|
|
} |
332
|
|
|
|