SwiftypeFileCrawlerExtensionTest::testUrlsNotToCrawlUnpublished() - 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 04:08 UTC

testUrlsNotToCrawlUnpublished() A

↳ Parent: SwiftypeFileCrawlerExtensionTest

Complexity

Conditions	1
Paths	1

Size

Total Lines	27
Code Lines	12

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
eloc	12
c	0
b	0
f	0
nc	1
nop	0
dl	0
loc	27
rs	9.8666

<?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();

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

        // 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();

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

        // Make sure we don't have any Cache set from the above publishing
        $this->assertEquals($urls, $file->getUrlsToCrawl());

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

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

    /**
     * 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();

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

        $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			// Make sure we don't have any Cache set from the above publishing
76			$file->clearCacheAll();
77
78			// Grab the Urls that we expect to have been collated
79			$key = str_replace('\\', '', $file->ClassName . $file->ID);
80			$urlsToCrawl = $file->getUrlsToCrawl();
81
82			// Check that the key exists for our file
83			$this->assertArrayHasKey($key, $urlsToCrawl);
84
85			// Grab the Urls that are for our page
86			$urlsToCrawl = $urlsToCrawl[$key];
87
88			// Strip out any http/https stuff
89			foreach ($urlsToCrawl as $urlToCrawl) {
90			$url = str_replace('http://', '', $urlToCrawl);
91			$url = str_replace('https://', '', $url);
92
93			$urls[] = $url;
94			}
95
96			$this->assertEquals($this->expectedUrls, $urls, '', 0.0, 10, true);
97			}
98
99			/**
100			* @throws Exception
101			*/
102			public function testUrlsToCrawlUnPublished(): void
103			{
104			/** @var SwiftypeFile $file */
105			$file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
106			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
107			$file->setFromLocalFile($sourcePath, $file->Filename);
108
109			// check our urls are not populated before we publish.
110			$urls = [];
111			$this->assertEquals($urls, $file->getUrlsToCrawl());
112
113			// Make sure our file is published before we begin
114			$file->publishSingle();
115
116			// Make sure we don't have any Cache set from the above publishing
117			$file->clearCacheAll();
118
119			// now we call the service to remove this file from index
120			$file->doUnpublish();
121
122			// Grab the Urls that we expect to have been collated
123			$key = str_replace('\\', '', $file->ClassName . $file->ID);
124			$urlsToCrawl = $file->getUrlsToCrawl();
125
126			// Check that the key exists for our page
127			$this->assertArrayHasKey($key, $urlsToCrawl);
128
129			// Grab the Urls that are for our page
130			$urlsToCrawl = $urlsToCrawl[$key];
131
132			// Strip out any http/https stuff
133			foreach ($urlsToCrawl as $urlToCrawl) {
134			$url = str_replace('http://', '', $urlToCrawl);
135			$url = str_replace('https://', '', $url);
136
137			$urls[] = $url;
138			}
139
140			$this->assertEquals($this->expectedUrls, $urls, '', 0.0, 10, true);
141			}
142
143			/**
144			* Files that hold no text (e.g. images) should not be indexed
145			*
146			* @throws Exception
147			*/
148			public function testUrlsNotToCrawlPublished(): void
149			{
150			/** @var SwiftypeFile $file */
151			$file = $this->objFromFixture(SwiftypeFile::class, 'file_jpg');
152			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
153			$file->setFromLocalFile($sourcePath, $file->Filename);
154
155			// check our urls are not populated before we publish.
156			$urls = [];
157			$this->assertEquals($urls, $file->getUrlsToCrawl());
158
159			// Publish single so that Urls to crawl is populated
160			$file->publishSingle();
161
162			// Make sure we don't have any Cache set from the above publishing
163			$file->clearCacheAll();
164
165			// Make sure we don't have any Cache set from the above publishing
166			$this->assertEquals($urls, $file->getUrlsToCrawl());
167
168			// Grab the Urls that we expect to have been collated
169			$urlsToCrawl = $file->getUrlsToCrawl();
170			$this->assertEquals($urls, $file->getUrlsToCrawl());
171
172			// Check that the key does not exist for our File
173			$key = str_replace('\\', '', $file->ClassName . $file->ID);
174			$this->assertArrayNotHasKey($key, $urlsToCrawl);
175			}
176
177			/**
178			* Files that hold no text (e.g. images) should not be indexed
179			*
180			* @throws Exception
181			*/
182			public function testUrlsNotToCrawlUnpublished(): void
183			{
184			/** @var SwiftypeFile $file */
185			$file = $this->objFromFixture(SwiftypeFile::class, 'file_jpg');
186			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
187			$file->setFromLocalFile($sourcePath, $file->Filename);
188
189			// check our urls are not populated before we publish.
190			$urls = [];
191			$this->assertEquals($urls, $file->getUrlsToCrawl());
192
193			// Make sure our file is published before we begin
194			$file->publishSingle();
195
196			// Make sure we don't have any Cache set from the above publishing
197			$file->clearCacheAll();
198
199			// now we call the service to remove this file from index
200			$file->doUnpublish();
201
202			// Grab the Urls that we expect to have been collated
203			$urlsToCrawl = $file->getUrlsToCrawl();
204			$this->assertEquals($urlsToCrawl, $file->getUrlsToCrawl());
205
206			// Check that the key does not exist for our File
207			$key = str_replace('\\', '', $file->ClassName . $file->ID);
208			$this->assertArrayNotHasKey($key, $urlsToCrawl);
209			}
210
211			/**
212			* @throws Exception
213			*/
214			public function testUrlsToCrawlSegmentChanged(): void
215			{
216			/** @var SwiftypeFile $file */
217			$file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
218			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
219			$file->setFromLocalFile($sourcePath, $file->Filename);
220
221			// check our urls are not populated before we publish anything.
222			$urls = [];
223			$this->assertEquals($urls, $file->getUrlsToCrawl());
224
225			// Make sure our file is published before we begin
226			$file->publishSingle();
227
228			$key = str_replace('\\', '', $file->ClassName . $file->ID);
229
230			// Make sure our cache is flushed from the above publishing
231			$file->clearCacheAll();
232
233			/**
234			* Note:
235			* Somehow because File DBObject records in Stage.Stage create a hash path in the URL to access the file,
236			* the asserted array contained 3 elements instead of 2. To ignore this we only asserted
237			* that our old and new file URL's exist in the array that is sent for reindexing.
238			*/
239			// Update our URL Segment
240			Versioned::withVersionedMode(function () use ($file) {
241			Versioned::set_reading_mode('Stage.Stage');
242			$file->renameFile('dummy-new.pdf');
243			});
244
245			// publish our file again to get new URL.
246			$file->publishSingle();
247
248			// We expect two URL's now. One from before the file rename change, and one from after it
249			$this->expectedUrls = [
250			'old_file' => 'localhost/assets/SwiftypeFileCrawlerExtensionTest/dummy.pdf',
251			'new_file' => 'localhost/assets/SwiftypeFileCrawlerExtensionTest/dummy-new.pdf',
252			];
253
254			// Grab the Urls that we expect to have been collated
255			$urlsToCrawl = $file->getUrlsToCrawl();
256
257			// Check that the key exists for our page
258			$this->assertArrayHasKey($key, $urlsToCrawl);
259
260			// Grab the Urls that are for our file
261			$urlsToCrawl = $urlsToCrawl[$key];
262
263			// Strip out any http/https stuff
264			foreach ($urlsToCrawl as $urlToCrawl) {
265			$url = str_replace('http://', '', $urlToCrawl);
266			$url = str_replace('https://', '', $url);
267
268			$urls[] = $url;
269			}
270
271			$this->assertContains($this->expectedUrls['old_file'], $urls);
272			$this->assertContains($this->expectedUrls['new_file'], $urls);
273			}
274
275			public function testUrlsToCrawlCacheCleared(): void
276			{
277			// Since asserting cache is cleared we want to reenable cache for this test.
278			Config::inst()->update(
279			SwiftypeFileCrawlerExtension::class,
280			'clear_cache_disabled',
281			false
282			);
283
284			/** @var SwiftypeFile $file */
285			$file = $this->objFromFixture(SwiftypeFile::class, 'file_pdf');
286			$sourcePath = __DIR__ . '/../Fixtures/' . $file->Name;
287			$file->setFromLocalFile($sourcePath, $file->Filename);
288
289			// check our urls are not populated before we publish anything.
290			$urls = [];
291			$this->assertEquals($urls, $file->getUrlsToCrawl());
292
293			// Publish single so that Urls to crawl is populated
294			$file->publishSingle();
295
296			// Make sure we don't have any Cache set from the above publishing
297			$file->clearCacheAll();
298
299			$key = str_replace('\\', '', $file->ClassName . $file->ID);
300
301			// Grab the Urls that we expect to have been collated
302			$urlsToCrawl = $file->getUrlsToCrawl();
303
304			// Check that the key exists for our page
305			$this->assertArrayNotHasKey($key, $urlsToCrawl);
306			}
307
308			public function tearDown()
309			{
310			TestAssetStore::reset();
311			parent::tearDown();
312			}
313			}
314

ichaber / silverstripe-swiftype

Pull Request — master (#28)

testUrlsNotToCrawlUnpublished() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like