SwiftypeFileCrawlerExtensionTest::testUrlsNotToCrawlPublished() - Code Metrics - Inspection of "Feature/reindex crawler for files" - ichaber/silverstripe-swiftype - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#28)

unknown

created 2021-04-08 04:41 UTC

testUrlsNotToCrawlPublished() A

↳ Parent: SwiftypeFileCrawlerExtensionTest

Complexity

Conditions	1
Paths	1

Size

Total Lines	17
Code Lines	8

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	1
eloc	8
c	1
b	0
f	0
nc	1
nop	0
dl	0
loc	17
rs	10

<?php

namespace Ichaber\SSSwiftype\Tests\Extensions;

use Exception;
use Ichaber\SSSwiftype\Extensions\SwiftypeFileCrawlerExtension;
use Ichaber\SSSwiftype\Extensions\SwiftypeSiteTreeCrawlerExtension;
use Ichaber\SSSwiftype\Tests\Fake\SwiftypeFile;
use SilverStripe\Assets\Dev\TestAssetStore;
use SilverStripe\Core\Config\Config;
use SilverStripe\Core\Injector\Injector;
use SilverStripe\Dev\SapphireTest;
Use SilverStripe\Assets\File;
use SilverStripe\ORM\DataObject;
use SilverStripe\Versioned\Versioned;

/**
 * Class SwiftypeFileCrawlerExtensionTest
 *
 * @package Ichaber\SSSwiftype\Tests\Extensions
 */
class SwiftypeFileCrawlerExtensionTest extends SapphireTest
{
    /**
     * @var string
     */
    protected static $fixture_file = 'SwiftypeFileCrawlerExtensionTest.yml';

    /**
     * expected array of files to be indexed
     *
     * @var string[]
     */
    protected $expectedUrls = [
        'localhost/assets/SwiftypeFileCrawlerExtensionTest/dummy.pdf',
    ];

    public function setUp(): void
    {
        parent::setUp();

        // Make sure that our cache is cleared between tests
        /** @var SwiftypeFileCrawlerExtension $crawlerExtension */
        $crawlerExtension = Injector::inst()->get(SwiftypeFileCrawlerExtension::class);
        $crawlerExtension->clearCacheAll();

        // Set our config to not clear caches after un/publish, so that we can easily fetch the Urls for our test
        Config::inst()->update(
            SwiftypeFileCrawlerExtension::class,
            'clear_cache_disabled',
            true
        );

        // Set backend assets store root to /SwiftypeFileCrawlerExtensionTest
        TestAssetStore::activate('SwiftypeFileCrawlerExtensionTest');
    }

    /**
     * @throws Exception
     */
    public function testUrlsToCrawlPublished(): void
    {
        /** @var SwiftypeFile $file */
        $file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
        $sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
        $file->setFromLocalFile($sourcePath, $file->Filename);

        // check our urls are not populated before we publish.
        $urls = [];
        $this->assertEquals($urls, $file->getUrlsToCrawl());

        // Publish single so that Urls to crawl is populated
        $file->publishSingle();

        // Grab the Urls that we expect to have been collated
        $key = str_replace('\\', '', $file->ClassName . $file->ID);
        $urlsToCrawl = $file->getUrlsToCrawl();

        // Check that the key exists for our file
        $this->assertArrayHasKey($key, $urlsToCrawl);

        // Grab the Urls that are for our page
        $urlsToCrawl = $urlsToCrawl[$key];

        // Strip out any http/https stuff
        foreach ($urlsToCrawl as $urlToCrawl) {
            $url = str_replace('http://', '', $urlToCrawl);
            $url = str_replace('https://', '', $url);

            $urls[] = $url;
        }

        $this->assertEquals($this->expectedUrls, $urls, '', 0.0, 10, true);
    }

    /**
     * @throws Exception
     */
    public function testUrlsToCrawlUnPublished(): void
    {
        /** @var SwiftypeFile $file */
        $file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
        $sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
        $file->setFromLocalFile($sourcePath, $file->Filename);

        // check our urls are not populated before we publish.
        $urls = [];
        $this->assertEquals($urls, $file->getUrlsToCrawl());

        // Make sure our file is published before we begin
        $file->publishSingle();

        // Make sure we don't have any Cache set from the above publishing
        $file->clearCacheAll();

        // now we call the service to remove this file from index
        $file->doUnpublish();

        // Grab the Urls that we expect to have been collated
        $key = str_replace('\\', '', $file->ClassName . $file->ID);
        $urlsToCrawl = $file->getUrlsToCrawl();

        // Check that the key exists for our page
        $this->assertArrayHasKey($key, $urlsToCrawl);

        // Grab the Urls that are for our page
        $urlsToCrawl = $urlsToCrawl[$key];

        // Strip out any http/https stuff
        foreach ($urlsToCrawl as $urlToCrawl) {
            $url = str_replace('http://', '', $urlToCrawl);
            $url = str_replace('https://', '', $url);

            $urls[] = $url;
        }

        $this->assertEquals($this->expectedUrls, $urls, '', 0.0, 10, true);
    }

    /**
     * Files that hold no text (e.g. images) should not be indexed
     *
     * @throws Exception
     */
    public function testUrlsNotToCrawlPublished(): void
    {
        /** @var SwiftypeFile $file */
        $file = $this->objFromFixture(SwiftypeFile::class, 'file_jpg');
        $sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
        $file->setFromLocalFile($sourcePath, $file->Filename);

        // check our urls are not populated before we publish.
        $urls = [];
        $this->assertEquals($urls, $file->getUrlsToCrawl());

        // Publish single so that Urls to crawl is populated
        $file->publishSingle();

        // Grab the Urls that we expect to have been collated
        $urlsToCrawl = $file->getUrlsToCrawl();

        $this->assertEquals($urls, $file->getUrlsToCrawl());
    }

    /**
     * Files that hold no text (e.g. images) should not be indexed
     *
     * @throws Exception
     */
    public function testUrlsNotToCrawlUnpublished(): void
    {
        /** @var SwiftypeFile $file */
        $file = $this->objFromFixture(SwiftypeFile::class, 'file_jpg');
        $sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
        $file->setFromLocalFile($sourcePath, $file->Filename);

        // check our urls are not populated before we publish.
        $urls = [];
        $this->assertEquals($urls, $file->getUrlsToCrawl());

        // Make sure our file is published before we begin
        $file->publishSingle();

        // Make sure we don't have any Cache set from the above publishing
        $file->clearCacheAll();

        // now we call the service to remove this file from index
        $file->doUnpublish();

        // Grab the Urls that we expect to have been collated
        $urlsToCrawl = $file->getUrlsToCrawl();
        $this->assertEquals($urlsToCrawl, $file->getUrlsToCrawl());

        // Check that the key does not exist for our File
        $key = str_replace('\\', '', $file->ClassName . $file->ID);
        $this->assertArrayNotHasKey($key, $urlsToCrawl);
    }

    /**
     * @throws Exception
     */
    public function testUrlsToCrawlSegmentChanged(): void
    {
        /** @var SwiftypeFile $file */
        $file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
        $sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
        $file->setFromLocalFile($sourcePath, $file->Filename);

        // check our urls are not populated before we publish anything.
        $urls = [];
        $this->assertEquals($urls, $file->getUrlsToCrawl());

        // Make sure our file is published before we begin
        $file->publishSingle();

        $key = str_replace('\\', '', $file->ClassName . $file->ID);

        // Make sure our cache is flushed from the above publishing
        $file->clearCacheAll();

        /**
         * Note:
         *  Somehow because File DBObject records in Stage.Stage create a hash path in the URL to access the file,
         *  the asserted array contained 3 elements instead of 2. To ignore this we only asserted
         *  that our old and new file URL's exist in the array that is sent for reindexing.
         */
        // Update our URL Segment
        Versioned::withVersionedMode(function () use ($file) {
            Versioned::set_reading_mode('Stage.Stage');
            $file->renameFile('dummy-new.pdf');
        });

        // publish our file again to get new URL.
        $file->publishSingle();

        // We expect two URL's now. One from before the file rename change, and one from after it
        $this->expectedUrls = [
            'old_file' => 'localhost/assets/SwiftypeFileCrawlerExtensionTest/dummy.pdf',
            'new_file' => 'localhost/assets/SwiftypeFileCrawlerExtensionTest/dummy-new.pdf',
        ];

        // Grab the Urls that we expect to have been collated
        $urlsToCrawl = $file->getUrlsToCrawl();

        // Check that the key exists for our page
        $this->assertArrayHasKey($key, $urlsToCrawl);

        // Grab the Urls that are for our file
        $urlsToCrawl = $urlsToCrawl[$key];

        // Strip out any http/https stuff
        foreach ($urlsToCrawl as $urlToCrawl) {
            $url = str_replace('http://', '', $urlToCrawl);
            $url = str_replace('https://', '', $url);

            $urls[] = $url;
        }

        $this->assertContains($this->expectedUrls['old_file'], $urls);
        $this->assertContains($this->expectedUrls['new_file'], $urls);
    }

    public function testUrlsToCrawlCacheCleared(): void
    {
        // Since asserting cache is cleared we want to reenable cache for this test.
        Config::inst()->update(
            SwiftypeFileCrawlerExtension::class,
            'clear_cache_disabled',
            false
        );

        /** @var SwiftypeFile $file */
        $file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
        $sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
        $file->setFromLocalFile($sourcePath, $file->Filename);

        // check our urls are not populated before we publish anything.
        $urls = [];
        $this->assertEquals($urls, $file->getUrlsToCrawl());

        // Publish single so that Urls to crawl is populated
        $file->publishSingle();

        $key = str_replace('\\', '', $file->ClassName . $file->ID);

        // Grab the Urls that we expect to have been collated
        $urlsToCrawl = $file->getUrlsToCrawl();

        // Check that the key exists for our page
        $this->assertArrayNotHasKey($key, $urlsToCrawl);
    }

    public function tearDown()
    {
        TestAssetStore::reset();
        parent::tearDown();
    }
}


1			<?php
2
3			namespace Ichaber\SSSwiftype\Tests\Extensions;
4
5			use Exception;
6			use Ichaber\SSSwiftype\Extensions\SwiftypeFileCrawlerExtension;
7			use Ichaber\SSSwiftype\Extensions\SwiftypeSiteTreeCrawlerExtension;
8			use Ichaber\SSSwiftype\Tests\Fake\SwiftypeFile;
9			use SilverStripe\Assets\Dev\TestAssetStore;
10			use SilverStripe\Core\Config\Config;
11			use SilverStripe\Core\Injector\Injector;
12			use SilverStripe\Dev\SapphireTest;
13			Use SilverStripe\Assets\File;
14			use SilverStripe\ORM\DataObject;
15			use SilverStripe\Versioned\Versioned;
16
17			/**
18			* Class SwiftypeFileCrawlerExtensionTest
19			*
20			* @package Ichaber\SSSwiftype\Tests\Extensions
21			*/
22			class SwiftypeFileCrawlerExtensionTest extends SapphireTest
23			{
24			/**
25			* @var string
26			*/
27			protected static $fixture_file = 'SwiftypeFileCrawlerExtensionTest.yml';
28
29			/**
30			* expected array of files to be indexed
31			*
32			* @var string[]
33			*/
34			protected $expectedUrls = [
35			'localhost/assets/SwiftypeFileCrawlerExtensionTest/dummy.pdf',
36			];
37
38			public function setUp(): void
39			{
40			parent::setUp();
41
42			// Make sure that our cache is cleared between tests
43			/** @var SwiftypeFileCrawlerExtension $crawlerExtension */
44			$crawlerExtension = Injector::inst()->get(SwiftypeFileCrawlerExtension::class);
45			$crawlerExtension->clearCacheAll();
46
47			// Set our config to not clear caches after un/publish, so that we can easily fetch the Urls for our test
48			Config::inst()->update(
49			SwiftypeFileCrawlerExtension::class,
50			'clear_cache_disabled',
51			true
52			);
53
54			// Set backend assets store root to /SwiftypeFileCrawlerExtensionTest
55			TestAssetStore::activate('SwiftypeFileCrawlerExtensionTest');
56			}
57
58			/**
59			* @throws Exception
60			*/
61			public function testUrlsToCrawlPublished(): void
62			{
63			/** @var SwiftypeFile $file */
64			$file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
65			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
66			$file->setFromLocalFile($sourcePath, $file->Filename);
67
68			// check our urls are not populated before we publish.
69			$urls = [];
70			$this->assertEquals($urls, $file->getUrlsToCrawl());
71
72			// Publish single so that Urls to crawl is populated
73			$file->publishSingle();
74
75			// Grab the Urls that we expect to have been collated
76			$key = str_replace('\\', '', $file->ClassName . $file->ID);
77			$urlsToCrawl = $file->getUrlsToCrawl();
78
79			// Check that the key exists for our file
80			$this->assertArrayHasKey($key, $urlsToCrawl);
81
82			// Grab the Urls that are for our page
83			$urlsToCrawl = $urlsToCrawl[$key];
84
85			// Strip out any http/https stuff
86			foreach ($urlsToCrawl as $urlToCrawl) {
87			$url = str_replace('http://', '', $urlToCrawl);
88			$url = str_replace('https://', '', $url);
89
90			$urls[] = $url;
91			}
92
93			$this->assertEquals($this->expectedUrls, $urls, '', 0.0, 10, true);
94			}
95
96			/**
97			* @throws Exception
98			*/
99			public function testUrlsToCrawlUnPublished(): void
100			{
101			/** @var SwiftypeFile $file */
102			$file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
103			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
104			$file->setFromLocalFile($sourcePath, $file->Filename);
105
106			// check our urls are not populated before we publish.
107			$urls = [];
108			$this->assertEquals($urls, $file->getUrlsToCrawl());
109
110			// Make sure our file is published before we begin
111			$file->publishSingle();
112
113			// Make sure we don't have any Cache set from the above publishing
114			$file->clearCacheAll();
115
116			// now we call the service to remove this file from index
117			$file->doUnpublish();
118
119			// Grab the Urls that we expect to have been collated
120			$key = str_replace('\\', '', $file->ClassName . $file->ID);
121			$urlsToCrawl = $file->getUrlsToCrawl();
122
123			// Check that the key exists for our page
124			$this->assertArrayHasKey($key, $urlsToCrawl);
125
126			// Grab the Urls that are for our page
127			$urlsToCrawl = $urlsToCrawl[$key];
128
129			// Strip out any http/https stuff
130			foreach ($urlsToCrawl as $urlToCrawl) {
131			$url = str_replace('http://', '', $urlToCrawl);
132			$url = str_replace('https://', '', $url);
133
134			$urls[] = $url;
135			}
136
137			$this->assertEquals($this->expectedUrls, $urls, '', 0.0, 10, true);
138			}
139
140			/**
141			* Files that hold no text (e.g. images) should not be indexed
142			*
143			* @throws Exception
144			*/
145			public function testUrlsNotToCrawlPublished(): void
146			{
147			/** @var SwiftypeFile $file */
148			$file = $this->objFromFixture(SwiftypeFile::class, 'file_jpg');
149			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
150			$file->setFromLocalFile($sourcePath, $file->Filename);
151
152			// check our urls are not populated before we publish.
153			$urls = [];
154			$this->assertEquals($urls, $file->getUrlsToCrawl());
155
156			// Publish single so that Urls to crawl is populated
157			$file->publishSingle();
158
159			// Grab the Urls that we expect to have been collated
160			$urlsToCrawl = $file->getUrlsToCrawl();
			0 ignored issues – show Unused Code introduced 2021-04-08 04:43 UTC by Report Bug Copy Issue Report The assignment to `$urlsToCrawl` is dead and can be removed. Loading history...
161			$this->assertEquals($urls, $file->getUrlsToCrawl());
162			}
163
164			/**
165			* Files that hold no text (e.g. images) should not be indexed
166			*
167			* @throws Exception
168			*/
169			public function testUrlsNotToCrawlUnpublished(): void
170			{
171			/** @var SwiftypeFile $file */
172			$file = $this->objFromFixture(SwiftypeFile::class, 'file_jpg');
173			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
174			$file->setFromLocalFile($sourcePath, $file->Filename);
175
176			// check our urls are not populated before we publish.
177			$urls = [];
178			$this->assertEquals($urls, $file->getUrlsToCrawl());
179
180			// Make sure our file is published before we begin
181			$file->publishSingle();
182
183			// Make sure we don't have any Cache set from the above publishing
184			$file->clearCacheAll();
185
186			// now we call the service to remove this file from index
187			$file->doUnpublish();
188
189			// Grab the Urls that we expect to have been collated
190			$urlsToCrawl = $file->getUrlsToCrawl();
191			$this->assertEquals($urlsToCrawl, $file->getUrlsToCrawl());
192
193			// Check that the key does not exist for our File
194			$key = str_replace('\\', '', $file->ClassName . $file->ID);
195			$this->assertArrayNotHasKey($key, $urlsToCrawl);
196			}
197
198			/**
199			* @throws Exception
200			*/
201			public function testUrlsToCrawlSegmentChanged(): void
202			{
203			/** @var SwiftypeFile $file */
204			$file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
205			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
206			$file->setFromLocalFile($sourcePath, $file->Filename);
207
208			// check our urls are not populated before we publish anything.
209			$urls = [];
210			$this->assertEquals($urls, $file->getUrlsToCrawl());
211
212			// Make sure our file is published before we begin
213			$file->publishSingle();
214
215			$key = str_replace('\\', '', $file->ClassName . $file->ID);
216
217			// Make sure our cache is flushed from the above publishing
218			$file->clearCacheAll();
219
220			/**
221			* Note:
222			* Somehow because File DBObject records in Stage.Stage create a hash path in the URL to access the file,
223			* the asserted array contained 3 elements instead of 2. To ignore this we only asserted
224			* that our old and new file URL's exist in the array that is sent for reindexing.
225			*/
226			// Update our URL Segment
227			Versioned::withVersionedMode(function () use ($file) {
228			Versioned::set_reading_mode('Stage.Stage');
229			$file->renameFile('dummy-new.pdf');
230			});
231
232			// publish our file again to get new URL.
233			$file->publishSingle();
234
235			// We expect two URL's now. One from before the file rename change, and one from after it
236			$this->expectedUrls = [
237			'old_file' => 'localhost/assets/SwiftypeFileCrawlerExtensionTest/dummy.pdf',
238			'new_file' => 'localhost/assets/SwiftypeFileCrawlerExtensionTest/dummy-new.pdf',
239			];
240
241			// Grab the Urls that we expect to have been collated
242			$urlsToCrawl = $file->getUrlsToCrawl();
243
244			// Check that the key exists for our page
245			$this->assertArrayHasKey($key, $urlsToCrawl);
246
247			// Grab the Urls that are for our file
248			$urlsToCrawl = $urlsToCrawl[$key];
249
250			// Strip out any http/https stuff
251			foreach ($urlsToCrawl as $urlToCrawl) {
252			$url = str_replace('http://', '', $urlToCrawl);
253			$url = str_replace('https://', '', $url);
254
255			$urls[] = $url;
256			}
257
258			$this->assertContains($this->expectedUrls['old_file'], $urls);
259			$this->assertContains($this->expectedUrls['new_file'], $urls);
260			}
261
262			public function testUrlsToCrawlCacheCleared(): void
263			{
264			// Since asserting cache is cleared we want to reenable cache for this test.
265			Config::inst()->update(
266			SwiftypeFileCrawlerExtension::class,
267			'clear_cache_disabled',
268			false
269			);
270
271			/** @var SwiftypeFile $file */
272			$file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
273			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
274			$file->setFromLocalFile($sourcePath, $file->Filename);
275
276			// check our urls are not populated before we publish anything.
277			$urls = [];
278			$this->assertEquals($urls, $file->getUrlsToCrawl());
279
280			// Publish single so that Urls to crawl is populated
281			$file->publishSingle();
282
283			$key = str_replace('\\', '', $file->ClassName . $file->ID);
284
285			// Grab the Urls that we expect to have been collated
286			$urlsToCrawl = $file->getUrlsToCrawl();
287
288			// Check that the key exists for our page
289			$this->assertArrayNotHasKey($key, $urlsToCrawl);
290			}
291
292			public function tearDown()
293			{
294			TestAssetStore::reset();
295			parent::tearDown();
296			}
297			}
298

ichaber / silverstripe-swiftype

Pull Request — master (#28)

testUrlsNotToCrawlPublished() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like