|
1
|
|
|
<?php declare(strict_types=1); |
|
2
|
|
|
|
|
3
|
|
|
namespace MOrtola\BehatSEOContexts\Context; |
|
4
|
|
|
|
|
5
|
|
|
use Behat\Mink\Exception\DriverException; |
|
6
|
|
|
use DOMDocument; |
|
7
|
|
|
use DOMElement; |
|
8
|
|
|
use DOMNodeList; |
|
9
|
|
|
use DOMXPath; |
|
10
|
|
|
use InvalidArgumentException; |
|
11
|
|
|
use MOrtola\BehatSEOContexts\Exception\InvalidOrderException; |
|
12
|
|
|
use Symfony\Component\Routing\Exception\RouteNotFoundException; |
|
13
|
|
|
use Webmozart\Assert\Assert; |
|
14
|
|
|
|
|
15
|
|
|
class SitemapContext extends BaseContext |
|
16
|
|
|
{ |
|
17
|
|
|
const SITEMAP_SCHEMA_FILE = __DIR__ . '/../Resources/schemas/sitemap.xsd'; |
|
18
|
|
|
const SITEMAP_XHTML_SCHEMA_FILE = __DIR__ . '/../Resources/schemas/sitemap_xhtml.xsd'; |
|
19
|
|
|
const SITEMAP_INDEX_SCHEMA_FILE = __DIR__ . '/../Resources/schemas/sitemap_index.xsd'; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* @var DOMDocument |
|
23
|
|
|
*/ |
|
24
|
|
|
private $sitemapXml; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @Given the sitemap :sitemapUrl |
|
28
|
|
|
*/ |
|
29
|
|
|
public function theSitemap(string $sitemapUrl): void |
|
30
|
|
|
{ |
|
31
|
|
|
$this->sitemapXml = $this->getSitemapXml($sitemapUrl); |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
private function getSitemapXml(string $sitemapUrl): DOMDocument |
|
35
|
|
|
{ |
|
36
|
|
|
$xml = new DOMDocument(); |
|
37
|
|
|
@$xmlLoaded = $xml->load($this->toAbsoluteUrl($sitemapUrl)); |
|
38
|
|
|
|
|
39
|
|
|
Assert::true($xmlLoaded, 'Error loading %s Sitemap using DOMDocument'); |
|
40
|
|
|
|
|
41
|
|
|
return $xml; |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* @throws InvalidOrderException |
|
46
|
|
|
* |
|
47
|
|
|
* @Then the index sitemap should have a child with URL :childSitemapUrl |
|
48
|
|
|
*/ |
|
49
|
|
|
public function theIndexSitemapShouldHaveAChildWithUrl(string $childSitemapUrl): void |
|
50
|
|
|
{ |
|
51
|
|
|
$this->assertSitemapHasBeenRead(); |
|
52
|
|
|
|
|
53
|
|
|
$xpathExpression = sprintf( |
|
54
|
|
|
'//sm:sitemapindex/sm:sitemap/sm:loc[contains(text(),"%s")]', |
|
55
|
|
|
$childSitemapUrl |
|
56
|
|
|
); |
|
57
|
|
|
|
|
58
|
|
|
Assert::greaterThanEq( |
|
59
|
|
|
$this->getXpathInspector()->query($xpathExpression)->length, |
|
60
|
|
|
1, |
|
61
|
|
|
sprintf( |
|
62
|
|
|
'Sitemap index %s has not child sitemap %s', |
|
63
|
|
|
$this->sitemapXml->documentURI, |
|
64
|
|
|
$childSitemapUrl |
|
65
|
|
|
) |
|
66
|
|
|
); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* @throws InvalidOrderException |
|
71
|
|
|
*/ |
|
72
|
|
|
private function assertSitemapHasBeenRead(): void |
|
73
|
|
|
{ |
|
74
|
|
|
if (!isset($this->sitemapXml)) { |
|
75
|
|
|
throw new InvalidOrderException( |
|
76
|
|
|
'You should execute "Given the sitemap :sitemapUrl" step before executing this step.' |
|
77
|
|
|
); |
|
78
|
|
|
} |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
private function getXpathInspector(): DOMXPath |
|
82
|
|
|
{ |
|
83
|
|
|
$xpath = new DOMXPath($this->sitemapXml); |
|
84
|
|
|
$xpath->registerNamespace('sm', 'http://www.sitemaps.org/schemas/sitemap/0.9'); |
|
85
|
|
|
$xpath->registerNamespace('xhtml', 'http://www.w3.org/1999/xhtml'); |
|
86
|
|
|
|
|
87
|
|
|
return $xpath; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* @throws InvalidOrderException |
|
92
|
|
|
* |
|
93
|
|
|
* @Then /^the sitemap should have ([0-9]+) children$/ |
|
94
|
|
|
*/ |
|
95
|
|
|
public function theSitemapShouldHaveChildren(int $expectedChildrenCount): void |
|
96
|
|
|
{ |
|
97
|
|
|
$this->assertSitemapHasBeenRead(); |
|
98
|
|
|
|
|
99
|
|
|
$sitemapChildrenCount = $this |
|
100
|
|
|
->getXpathInspector() |
|
101
|
|
|
->query('/*[self::sm:sitemapindex or self::sm:urlset]/*[self::sm:sitemap or self::sm:url]/sm:loc') |
|
102
|
|
|
->length; |
|
103
|
|
|
|
|
104
|
|
|
Assert::eq( |
|
105
|
|
|
$expectedChildrenCount, |
|
106
|
|
|
$sitemapChildrenCount, |
|
107
|
|
|
sprintf( |
|
108
|
|
|
'Sitemap %s has %d children, expected value was: %d', |
|
109
|
|
|
$this->sitemapXml->documentURI, |
|
110
|
|
|
$sitemapChildrenCount, |
|
111
|
|
|
$expectedChildrenCount |
|
112
|
|
|
) |
|
113
|
|
|
); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* @throws InvalidOrderException |
|
118
|
|
|
* |
|
119
|
|
|
* @Then the multilanguage sitemap should pass Google validation |
|
120
|
|
|
*/ |
|
121
|
|
|
public function theMultilanguageSitemapShouldPassGoogleValidation(): void |
|
122
|
|
|
{ |
|
123
|
|
|
$this->assertSitemapHasBeenRead(); |
|
124
|
|
|
|
|
125
|
|
|
$this->assertValidSitemap(self::SITEMAP_XHTML_SCHEMA_FILE); |
|
126
|
|
|
|
|
127
|
|
|
$urlsNodes = $this->getXpathInspector()->query('//sm:urlset/sm:url'); |
|
128
|
|
|
|
|
129
|
|
|
/** @var DOMElement $urlNode */ |
|
130
|
|
|
foreach ($urlsNodes as $urlNode) { |
|
131
|
|
|
$urlElement = $urlNode->getElementsByTagName('loc')->item(0); |
|
132
|
|
|
|
|
133
|
|
|
Assert::notNull($urlElement); |
|
134
|
|
|
|
|
135
|
|
|
$urlLoc = $urlElement->nodeValue; |
|
136
|
|
|
|
|
137
|
|
|
/** @var DOMElement $alternateLink */ |
|
138
|
|
|
foreach ($urlNode->getElementsByTagName('link') as $alternateLink) { |
|
139
|
|
|
$alternateLinkHref = $alternateLink->getAttribute('href'); |
|
140
|
|
|
|
|
141
|
|
|
if ($alternateLinkHref !== $urlLoc) { |
|
142
|
|
|
$alternateLinkNodes = $this->getXpathInspector()->query( |
|
143
|
|
|
sprintf('//sm:urlset/sm:url/sm:loc[text()="%s"]', $alternateLinkHref) |
|
144
|
|
|
); |
|
145
|
|
|
|
|
146
|
|
|
Assert::greaterThanEq( |
|
147
|
|
|
$alternateLinkNodes->length, |
|
148
|
|
|
1, |
|
149
|
|
|
sprintf( |
|
150
|
|
|
'Url %s has not reciprocous URL for alternative link %s in Sitemap %s', |
|
151
|
|
|
$urlLoc, |
|
152
|
|
|
$alternateLinkHref, |
|
153
|
|
|
$this->sitemapXml->documentURI |
|
154
|
|
|
) |
|
155
|
|
|
); |
|
156
|
|
|
} |
|
157
|
|
|
} |
|
158
|
|
|
} |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
private function assertValidSitemap(string $sitemapSchemaFile): void |
|
162
|
|
|
{ |
|
163
|
|
|
Assert::fileExists( |
|
164
|
|
|
$sitemapSchemaFile, |
|
165
|
|
|
sprintf('Sitemap schema file %s does not exist', $sitemapSchemaFile) |
|
166
|
|
|
); |
|
167
|
|
|
|
|
168
|
|
|
Assert::true( |
|
169
|
|
|
@$this->sitemapXml->schemaValidate($sitemapSchemaFile), |
|
170
|
|
|
sprintf( |
|
171
|
|
|
'Sitemap %s does not pass validation using %s schema', |
|
172
|
|
|
$this->sitemapXml->documentURI, |
|
173
|
|
|
$sitemapSchemaFile |
|
174
|
|
|
) |
|
175
|
|
|
); |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
/** |
|
179
|
|
|
* @throws InvalidOrderException |
|
180
|
|
|
* @throws DriverException |
|
181
|
|
|
* |
|
182
|
|
|
* @Then the sitemap URLs should be alive |
|
183
|
|
|
*/ |
|
184
|
|
|
public function theSitemapUrlsShouldBeAlive(): void |
|
185
|
|
|
{ |
|
186
|
|
|
$this->assertSitemapHasBeenRead(); |
|
187
|
|
|
|
|
188
|
|
|
$locNodes = $this->getXpathInspector()->query('//sm:urlset/sm:url/sm:loc'); |
|
189
|
|
|
|
|
190
|
|
|
Assert::isInstanceOf($locNodes, DOMNodeList::class); |
|
191
|
|
|
|
|
192
|
|
|
foreach ($locNodes as $locNode) { |
|
193
|
|
|
$this->urlIsValid($locNode); |
|
194
|
|
|
$this->urlIsAlive($locNode); |
|
195
|
|
|
} |
|
196
|
|
|
} |
|
197
|
|
|
|
|
198
|
|
|
/** |
|
199
|
|
|
* @throws DriverException |
|
200
|
|
|
*/ |
|
201
|
|
|
private function urlIsValid($locNode): void |
|
202
|
|
|
{ |
|
203
|
|
|
try { |
|
204
|
|
|
$this->visit($locNode->nodeValue); |
|
205
|
|
|
} catch (RouteNotFoundException $e) { |
|
206
|
|
|
throw new InvalidArgumentException( |
|
207
|
|
|
sprintf( |
|
208
|
|
|
'Sitemap Url %s is not valid in Sitemap: %s. Exception: %s', |
|
209
|
|
|
$locNode->nodeValue, |
|
210
|
|
|
$this->sitemapXml->documentURI, |
|
211
|
|
|
$e->getMessage() |
|
212
|
|
|
), |
|
213
|
|
|
0, |
|
214
|
|
|
$e |
|
215
|
|
|
); |
|
216
|
|
|
} |
|
217
|
|
|
} |
|
218
|
|
|
|
|
219
|
|
|
private function urlIsAlive($locNode): void |
|
220
|
|
|
{ |
|
221
|
|
|
Assert::eq( |
|
222
|
|
|
200, |
|
223
|
|
|
$this->getStatusCode(), |
|
224
|
|
|
sprintf( |
|
225
|
|
|
'Sitemap Url %s is not valid in Sitemap: %s. Response status code: %s', |
|
226
|
|
|
$locNode->nodeValue, |
|
227
|
|
|
$this->sitemapXml->documentURI, |
|
228
|
|
|
$this->getStatusCode() |
|
229
|
|
|
) |
|
230
|
|
|
); |
|
231
|
|
|
} |
|
232
|
|
|
|
|
233
|
|
|
/** |
|
234
|
|
|
* @throws DriverException |
|
235
|
|
|
* @throws InvalidOrderException |
|
236
|
|
|
* |
|
237
|
|
|
* @Then :randomUrlsCount random sitemap URL/URLs should be alive |
|
238
|
|
|
*/ |
|
239
|
|
|
public function randomSitemapUrlsShouldBeAlive(int $randomUrlsCount): void |
|
240
|
|
|
{ |
|
241
|
|
|
$this->assertSitemapHasBeenRead(); |
|
242
|
|
|
|
|
243
|
|
|
$locNodes = $this->getXpathInspector()->query('//sm:urlset/sm:url/sm:loc'); |
|
244
|
|
|
|
|
245
|
|
|
Assert::isInstanceOf($locNodes, DOMNodeList::class); |
|
246
|
|
|
|
|
247
|
|
|
$locNodesCount = $locNodes->length; |
|
248
|
|
|
|
|
249
|
|
|
Assert::greaterThan( |
|
250
|
|
|
$locNodesCount, |
|
251
|
|
|
$randomUrlsCount, |
|
252
|
|
|
sprintf( |
|
253
|
|
|
'Sitemap %s only has %d children, minimum expected value was: %d', |
|
254
|
|
|
$this->sitemapXml->documentURI, |
|
255
|
|
|
$locNodesCount, |
|
256
|
|
|
$randomUrlsCount |
|
257
|
|
|
) |
|
258
|
|
|
); |
|
259
|
|
|
|
|
260
|
|
|
$locNodesArray = iterator_to_array($locNodes); |
|
|
|
|
|
|
261
|
|
|
shuffle($locNodesArray); |
|
262
|
|
|
for ($i = 0; $i <= $randomUrlsCount - 1; $i++) { |
|
263
|
|
|
$this->urlIsValid($locNodesArray[$i]); |
|
264
|
|
|
$this->urlIsAlive($locNodesArray[$i]); |
|
265
|
|
|
} |
|
266
|
|
|
} |
|
267
|
|
|
|
|
268
|
|
|
/** |
|
269
|
|
|
* @Then /^the (index |multilanguage |)sitemap should not be valid$/ |
|
270
|
|
|
*/ |
|
271
|
|
|
public function theSitemapShouldNotBeValid(string $sitemapType = ''): void |
|
272
|
|
|
{ |
|
273
|
|
|
$this->assertInverse( |
|
274
|
|
|
function () use ($sitemapType) { |
|
275
|
|
|
$this->theSitemapShouldBeValid($sitemapType); |
|
276
|
|
|
}, |
|
277
|
|
|
sprintf('The sitemap is a valid %s sitemap.', $sitemapType) |
|
278
|
|
|
); |
|
279
|
|
|
} |
|
280
|
|
|
|
|
281
|
|
|
/** |
|
282
|
|
|
* @throws InvalidOrderException |
|
283
|
|
|
* |
|
284
|
|
|
* @Then /^the (index |multilanguage |)sitemap should be valid$/ |
|
285
|
|
|
*/ |
|
286
|
|
|
public function theSitemapShouldBeValid(string $sitemapType = ''): void |
|
287
|
|
|
{ |
|
288
|
|
|
$this->assertSitemapHasBeenRead(); |
|
289
|
|
|
|
|
290
|
|
|
switch (trim($sitemapType)) { |
|
291
|
|
|
case 'index': |
|
292
|
|
|
$sitemapSchemaFile = self::SITEMAP_INDEX_SCHEMA_FILE; |
|
293
|
|
|
|
|
294
|
|
|
break; |
|
295
|
|
|
case 'multilanguage': |
|
296
|
|
|
$sitemapSchemaFile = self::SITEMAP_XHTML_SCHEMA_FILE; |
|
297
|
|
|
|
|
298
|
|
|
break; |
|
299
|
|
|
default: |
|
300
|
|
|
$sitemapSchemaFile = self::SITEMAP_SCHEMA_FILE; |
|
301
|
|
|
} |
|
302
|
|
|
|
|
303
|
|
|
$this->assertValidSitemap($sitemapSchemaFile); |
|
304
|
|
|
} |
|
305
|
|
|
|
|
306
|
|
|
/** |
|
307
|
|
|
* @Then the multilanguage sitemap should not pass Google validation |
|
308
|
|
|
*/ |
|
309
|
|
|
public function theMultilanguageSitemapShouldNotPassGoogleValidation(): void |
|
310
|
|
|
{ |
|
311
|
|
|
$this->assertInverse( |
|
312
|
|
|
[$this, 'theMultilanguageSitemapShouldPassGoogleValidation'], |
|
313
|
|
|
sprintf('The multilanguage sitemap passes Google validation.') |
|
314
|
|
|
); |
|
315
|
|
|
} |
|
316
|
|
|
} |
|
317
|
|
|
|