|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace SiteChecker; |
|
4
|
|
|
|
|
5
|
|
|
use GuzzleHttp\Client; |
|
6
|
|
|
use GuzzleHttp\Cookie\CookieJar; |
|
7
|
|
|
use GuzzleHttp\Cookie\SetCookie; |
|
8
|
|
|
use GuzzleHttp\Exception\RequestException; |
|
9
|
|
|
use GuzzleHttp\RequestOptions; |
|
10
|
|
|
use Psr\Http\Message\ResponseInterface; |
|
11
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* Class SiteChecker |
|
15
|
|
|
* @package SiteChecker |
|
16
|
|
|
*/ |
|
17
|
|
|
class SiteChecker |
|
18
|
|
|
{ |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var array |
|
22
|
|
|
*/ |
|
23
|
|
|
protected $messages = []; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @var Asset[] |
|
27
|
|
|
*/ |
|
28
|
|
|
protected $checkedAssets = []; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @var Asset |
|
32
|
|
|
*/ |
|
33
|
|
|
protected $basePage; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* @var SiteCheckObserver |
|
37
|
|
|
*/ |
|
38
|
|
|
protected $observer; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* @var Client |
|
42
|
|
|
*/ |
|
43
|
|
|
protected $client; |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* @var Config |
|
47
|
|
|
*/ |
|
48
|
|
|
protected $config; |
|
49
|
|
|
|
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* @return Config |
|
53
|
|
|
*/ |
|
54
|
|
|
public function getConfig() |
|
55
|
|
|
{ |
|
56
|
|
|
return $this->config; |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
/** |
|
60
|
|
|
* @param Config $config |
|
61
|
|
|
*/ |
|
62
|
|
|
public function setConfig($config) |
|
63
|
|
|
{ |
|
64
|
|
|
$this->config = $config; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* SiteChecker constructor. |
|
69
|
|
|
* @param \GuzzleHttp\Client $client |
|
70
|
|
|
* @param \SiteChecker\SiteCheckObserver|null $observer |
|
71
|
|
|
*/ |
|
72
|
|
|
public function __construct( |
|
73
|
|
|
Client $client, |
|
74
|
|
|
SiteCheckObserver $observer = null |
|
75
|
|
|
) { |
|
76
|
|
|
$this->client = $client; |
|
77
|
|
|
$this->observer = $observer ?: new DummyObserver(); |
|
78
|
|
|
$this->config = new Config(); |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* @param \SiteChecker\SiteCheckObserver $observer |
|
84
|
|
|
* @return static |
|
85
|
|
|
*/ |
|
86
|
|
|
public static function create(SiteCheckObserver $observer) |
|
87
|
|
|
{ |
|
88
|
|
|
$client = new Client([ |
|
89
|
|
|
RequestOptions::ALLOW_REDIRECTS => true, |
|
90
|
|
|
RequestOptions::COOKIES => true, |
|
91
|
|
|
RequestOptions::VERIFY => false, |
|
92
|
|
|
]); |
|
93
|
|
|
|
|
94
|
|
|
return new static($client, $observer); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
/** |
|
98
|
|
|
* Check the site for broken assets. |
|
99
|
|
|
* |
|
100
|
|
|
* @param string $baseUrl |
|
101
|
|
|
*/ |
|
102
|
|
|
public function check($baseUrl) |
|
103
|
|
|
{ |
|
104
|
|
|
if (!$baseUrl instanceof Asset) { |
|
105
|
|
|
$baseUrl = new Asset($baseUrl); |
|
106
|
|
|
} |
|
107
|
|
|
$this->messages = []; |
|
108
|
|
|
$this->basePage = $baseUrl; |
|
109
|
|
|
|
|
110
|
|
|
$this->checkAsset($baseUrl); |
|
111
|
|
|
foreach ($this->config->includedUrls as $includedUrl) { |
|
112
|
|
|
$asset = new Asset($includedUrl); |
|
113
|
|
|
$this->normalizeUrl($asset); |
|
114
|
|
|
if (!in_array($asset->getUrl(), $this->checkedAssets)) { |
|
115
|
|
|
$this->checkAsset($asset); |
|
116
|
|
|
} |
|
117
|
|
|
} |
|
118
|
|
|
$this->observer->receiveResults($this->checkedAssets); |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* @param Asset $asset |
|
124
|
|
|
*/ |
|
125
|
|
|
protected function checkAsset(Asset $asset) |
|
126
|
|
|
{ |
|
127
|
|
|
if (!$this->shouldBeChecked($asset)) { |
|
128
|
|
|
return; |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
if (!$this->observer->pageToCheck($asset)) { |
|
132
|
|
|
return; |
|
133
|
|
|
} |
|
134
|
|
|
$cookies = $this->config->getCookies(); |
|
135
|
|
|
|
|
136
|
|
|
foreach ($cookies as $key => $cookie) { |
|
137
|
|
|
$cookie['Domain'] = $this->basePage->host; |
|
138
|
|
|
$cookies[$key] = new SetCookie($cookie); |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
$jar = new CookieJar(false, $cookies); |
|
142
|
|
|
|
|
143
|
|
|
try { |
|
144
|
|
|
$response = $this->client->request('GET', $asset->getURL(), |
|
145
|
|
|
[ |
|
146
|
|
|
'cookies' => $jar |
|
147
|
|
|
]); |
|
148
|
|
|
} catch (RequestException $exception) { |
|
149
|
|
|
$response = $exception->getResponse(); |
|
150
|
|
|
$asset->setResponseCode('500'); |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
if ($response) { |
|
154
|
|
|
$asset->setResponseCode($response->getStatusCode()); |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
$this->observer->pageChecked($asset, $response); |
|
158
|
|
|
|
|
159
|
|
|
$this->checkedAssets[] = $asset; |
|
160
|
|
|
|
|
161
|
|
|
if (!$response) { |
|
162
|
|
|
return; |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
if (!$this->isExternal($asset) && $this->isHtmlPage($response)) { |
|
166
|
|
|
$this->checkAllAssets($response->getBody()->getContents(), $asset); |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
|
|
/** |
|
172
|
|
|
* @param Asset $asset |
|
173
|
|
|
* @return bool |
|
174
|
|
|
*/ |
|
175
|
|
|
protected function isExternal(Asset $asset) |
|
176
|
|
|
{ |
|
177
|
|
|
return $this->basePage->host !== $asset->host; |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
/** |
|
181
|
|
|
* Crawl all assets in the given html. |
|
182
|
|
|
* |
|
183
|
|
|
* @param string $html |
|
184
|
|
|
* @param Asset $parentAsset |
|
185
|
|
|
*/ |
|
186
|
|
|
protected function checkAllAssets($html, $parentAsset) |
|
187
|
|
|
{ |
|
188
|
|
|
$allAssets = $this->getAllAssets($html, $parentAsset); |
|
189
|
|
|
|
|
190
|
|
|
/** @var Asset $asset */ |
|
191
|
|
|
foreach ($allAssets as $asset) { |
|
192
|
|
|
if ($asset->isHttp()) { |
|
193
|
|
|
$this->normalizeUrl($asset); |
|
194
|
|
|
if ($this->shouldBeChecked($asset)) { |
|
195
|
|
|
$this->checkAsset($asset); |
|
196
|
|
|
} |
|
197
|
|
|
} |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
} |
|
201
|
|
|
|
|
202
|
|
|
/** |
|
203
|
|
|
* Crawl all assets in the given html. |
|
204
|
|
|
* |
|
205
|
|
|
* @param $html |
|
206
|
|
|
* @param $parentPage |
|
207
|
|
|
* @return array |
|
208
|
|
|
*/ |
|
209
|
|
|
protected function getAllAssets($html, $parentPage) |
|
210
|
|
|
{ |
|
211
|
|
|
$assets = []; |
|
212
|
|
|
|
|
213
|
|
|
$assetTypes = [ |
|
214
|
|
|
'checkImages' => [ |
|
215
|
|
|
'//img', |
|
216
|
|
|
'src', |
|
217
|
|
|
'image', |
|
218
|
|
|
], |
|
219
|
|
|
'checkJS' => [ |
|
220
|
|
|
'//script', |
|
221
|
|
|
'src', |
|
222
|
|
|
'js file', |
|
223
|
|
|
], |
|
224
|
|
|
'checkCSS' => [ |
|
225
|
|
|
'//link[@rel="stylesheet"]', |
|
226
|
|
|
'href', |
|
227
|
|
|
'image', |
|
228
|
|
|
], |
|
229
|
|
|
]; |
|
230
|
|
|
|
|
231
|
|
|
$assets = array_merge( |
|
232
|
|
|
$assets, |
|
233
|
|
|
$this->createAssetsFromDOMElements( |
|
234
|
|
|
$html, '//a', 'href', 'page', $parentPage |
|
235
|
|
|
) |
|
236
|
|
|
); |
|
237
|
|
|
|
|
238
|
|
|
foreach ($assetTypes as $configKey => $args) { |
|
239
|
|
|
array_unshift($args, $html); |
|
240
|
|
|
$args[] = $parentPage; |
|
241
|
|
|
$assets = array_merge( |
|
242
|
|
|
$assets, |
|
243
|
|
|
call_user_func_array( |
|
244
|
|
|
[$this, "createAssetsFromDOMElements"], |
|
245
|
|
|
$args |
|
246
|
|
|
) |
|
247
|
|
|
); |
|
248
|
|
|
} |
|
249
|
|
|
|
|
250
|
|
|
return $assets; |
|
251
|
|
|
} |
|
252
|
|
|
|
|
253
|
|
|
/** |
|
254
|
|
|
* @param $html |
|
255
|
|
|
* @param $selector |
|
256
|
|
|
* @param $urlAttribute |
|
257
|
|
|
* @param $type |
|
258
|
|
|
* @param $parentPage |
|
259
|
|
|
* @return array |
|
260
|
|
|
*/ |
|
261
|
|
|
protected function createAssetsFromDOMElements( |
|
262
|
|
|
$html, |
|
263
|
|
|
$selector, |
|
264
|
|
|
$urlAttribute, |
|
265
|
|
|
$type, |
|
266
|
|
|
$parentPage |
|
267
|
|
|
) { |
|
268
|
|
|
$assets = []; |
|
269
|
|
|
|
|
270
|
|
|
$crawler = new Crawler($html); |
|
271
|
|
|
$elements = $crawler->filterXpath($selector); |
|
272
|
|
|
|
|
273
|
|
|
/** @var \DOMElement $assetElement */ |
|
274
|
|
|
foreach ($elements as $element) { |
|
275
|
|
|
if (!empty($element->getAttribute($urlAttribute))) { |
|
276
|
|
|
$urlValue = $element->getAttribute($urlAttribute); |
|
277
|
|
|
if ($this->config->ignoreWhiteSpaces) { |
|
278
|
|
|
$urlValue = trim($urlValue); |
|
279
|
|
|
} |
|
280
|
|
|
|
|
281
|
|
|
$assets[] = new Asset( |
|
282
|
|
|
$urlValue, |
|
283
|
|
|
$parentPage, |
|
284
|
|
|
$element->ownerDocument->saveHTML($element), |
|
285
|
|
|
$type |
|
286
|
|
|
); |
|
287
|
|
|
} |
|
288
|
|
|
} |
|
289
|
|
|
|
|
290
|
|
|
return $assets; |
|
291
|
|
|
} |
|
292
|
|
|
|
|
293
|
|
|
/** |
|
294
|
|
|
* @param \Psr\Http\Message\ResponseInterface $response |
|
295
|
|
|
* @return bool |
|
296
|
|
|
*/ |
|
297
|
|
|
protected function isHtmlPage(ResponseInterface $response) |
|
298
|
|
|
{ |
|
299
|
|
|
foreach ($response->getHeader('content-type') as $header) { |
|
300
|
|
|
if (stristr($header, 'text/html') !== false) { |
|
301
|
|
|
return true; |
|
302
|
|
|
} |
|
303
|
|
|
} |
|
304
|
|
|
return false; |
|
305
|
|
|
} |
|
306
|
|
|
|
|
307
|
|
|
/** |
|
308
|
|
|
* @param Asset $asset |
|
309
|
|
|
* @return bool |
|
310
|
|
|
*/ |
|
311
|
|
|
protected function shouldBeChecked(Asset $asset) |
|
312
|
|
|
{ |
|
313
|
|
|
if (in_array($asset->getURL(), $this->config->excludedUrls)) { |
|
314
|
|
|
return false; |
|
315
|
|
|
} |
|
316
|
|
|
foreach ($this->config->excludedUrls as $excludedUrl) { |
|
317
|
|
|
if (preg_match('/' . $excludedUrl . '/i', $asset->getURL())) { |
|
318
|
|
|
return false; |
|
319
|
|
|
} |
|
320
|
|
|
} |
|
321
|
|
|
if (!$this->config->checkExternal && $this->isExternal($asset)) { |
|
322
|
|
|
return false; |
|
323
|
|
|
} |
|
324
|
|
|
return !in_array($asset->getUrl(), $this->checkedAssets); |
|
325
|
|
|
} |
|
326
|
|
|
|
|
327
|
|
|
/** |
|
328
|
|
|
* @param \SiteChecker\Asset $asset |
|
329
|
|
|
* @return bool |
|
330
|
|
|
*/ |
|
331
|
|
|
protected function isAlreadyChecked(Asset $asset) |
|
332
|
|
|
{ |
|
333
|
|
|
return in_array($asset->getURL(), $this->checkedAssets); |
|
334
|
|
|
} |
|
335
|
|
|
|
|
336
|
|
|
|
|
337
|
|
|
/** |
|
338
|
|
|
* Normalize the given url. |
|
339
|
|
|
* @param \SiteChecker\Asset $asset |
|
340
|
|
|
* @return $this |
|
341
|
|
|
*/ |
|
342
|
|
|
protected function normalizeUrl(Asset $asset) |
|
343
|
|
|
{ |
|
344
|
|
|
if ($asset->isRelative()) { |
|
345
|
|
|
|
|
346
|
|
|
$asset->setScheme($this->basePage->scheme) |
|
347
|
|
|
->setHost($this->basePage->host) |
|
348
|
|
|
->setPort($this->basePage->port); |
|
349
|
|
|
} |
|
350
|
|
|
|
|
351
|
|
|
if ($asset->isProtocolIndependent()) { |
|
352
|
|
|
$asset->setScheme($this->basePage->scheme); |
|
353
|
|
|
} |
|
354
|
|
|
|
|
355
|
|
|
return $asset->removeFragment(); |
|
356
|
|
|
} |
|
357
|
|
|
|
|
358
|
|
|
/** |
|
359
|
|
|
* @return \SiteChecker\Asset[] |
|
360
|
|
|
*/ |
|
361
|
|
|
public function getResults() |
|
362
|
|
|
{ |
|
363
|
|
|
return $this->checkedAssets; |
|
364
|
|
|
} |
|
365
|
|
|
|
|
366
|
|
|
} |
|
367
|
|
|
|