SwiftypeSiteTreeCrawlerExtension::setUrlsToCrawl() - Code Metrics - Inspection of "Add version context to crawl url" - ichaber/silverstripe-swiftype - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#16)

by Chris

created 2019-12-13 00:28 UTC

SwiftypeSiteTreeCrawlerExtension::setUrlsToCrawl() A

↳ Parent: SwiftypeSiteTreeCrawlerExtension

Complexity

Conditions	1
Paths	1

Size

Total Lines	2
Code Lines	1

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	1
eloc	1
c	1
b	0
f	0
nc	1
nop	1
dl	0
loc	2
rs	10

<?php

namespace Ichaber\SSSwiftype\Extensions;

use Ichaber\SSSwiftype\Service\SwiftypeCrawler;
use SilverStripe\CMS\Model\SiteTree;
use SilverStripe\CMS\Model\SiteTreeExtension;
use SilverStripe\Control\Director;
use SilverStripe\Core\Config\Config;
use SilverStripe\Versioned\Versioned;

/**
 * Class SwiftypeSiteTreeCrawlerExtension
 *
 * @package Ichaber\SSSwiftype\Extensions
 * @property SiteTree|$this $owner
 */
class SwiftypeSiteTreeCrawlerExtension extends SiteTreeExtension
{
    /**
     * Urls to crawl
     *
     * array keyed by getOwnerKey
     *
     * @var array
     */
    private $urlsToCrawl = [];

    /**
     * @param array $urls
     */
    public function setUrlsToCrawl(array $urls) {
        $this->urlsToCrawl = $urls;
    }

    /**
     * @return array
     */
    public function getUrlsToCrawl(): array
    {
        return $this->urlsToCrawl;
    }

    /**
     * We need to collate Urls before we write, just in case an author has changed the Page's Url Segment. If they
     * have, then we need to request Swiftype to reindex both the old Url (which should then be marked by Swiftype
     * as a 404), and the new Url
     */
    public function onBeforeWrite(): void
    {
        $this->collateUrls();
    }

    /**
     * After a publish has occurred, we can collate and process immediately (no need to split things out like during
     * an unpublish)
     *
     * @param SiteTree|mixed $original
     * @return void
     */
    public function onAfterPublish(&$original): void
    {
        $this->collateUrls();
        $this->processCollatedUrls();

        // Check to see if the clearing of cache has been disabled (useful for unit testing, or any other reason you
        // might have to disable it)
        $clearCacheDisabled = Config::inst()->get(static::class, 'clear_cache_disabled');

        if ($clearCacheDisabled) {
            return;
        }

        // It's important that we clear the cache after we have finished requesting reindex from Swiftype
        $this->clearCacheSingle();
    }

    /**
     * We need to collate the Urls to be purged *before* we complete the unpublish action (otherwise, the LIVE Urls
     * will no longer be available, since the page is now unpublished)
     */
    public function onBeforeUnpublish(): void
    {
        $this->collateUrls();
    }

    /**
     * After the unpublish has completed, we can now request Swiftype to reindex the Urls that we collated
     */
    public function onAfterUnpublish(): void
    {
        $this->processCollatedUrls();

        // Check to see if the clearing of cache has been disabled (useful for unit testing, or any other reason you
        // might have to disable it)
        $clearCacheDisabled = Config::inst()->get(static::class, 'clear_cache_disabled');

        if ($clearCacheDisabled) {
            return;
        }

        // It's important that we clear the cache after we have finished requesting reindex from Swiftype
        $this->clearCacheSingle();
    }

    /**
     * You may need to clear the cache at some point during your particular process
     *
     * Reset all Urls for any/all objects that might be in the cache (keeping in mind that Extensions are singleton,
     * so the UrlsToCache could be accessed via singleton and it could contain Urls for many owner objects)
     *
     * We don't use flushCache (which is called from DataObject) because this is called between write and un/publish,
     * and we need our cache to persist through these states
     */
    public function clearCacheAll(): void
    {
        $this->setUrlsToCrawl([]);
    }

    /**
     * You may need to clear the cache at some point during your particular process
     *
     * Reset only the Urls related to this particular owner object (keeping in mind that Extensions are singleton,
     * so the UrlsToCache could be accessed via singleton and it could contain Urls for many owner objects)
     *
     * We don't use flushCache (which is called from DataObject) because this is called between write and un/publish,
     * and we need our cache to persist through these states
     */
    public function clearCacheSingle(): void
    {
        $urls = $this->getUrlsToCrawl();
        $key = $this->getOwnerKey();

        // Nothing for us to do here
        if ($key === null) {
            return;
        }

        // Nothing for us to do here
        if (!array_key_exists($key, $urls)) {
            return;
        }

        // Remove this key and it's Urls
        unset($urls[$key]);

        $this->setUrlsToCrawl($urls);
    }

    /**
     * Collate Urls to crawl
     *
     * Extensions are singleton, so we use the owner key to make sure that we're only processing Urls directly related
     * to the desired record.
     *
     * You might need to collate more than one URL per Page (maybe you're using Fluent or another translation module).
     * This is the method you will want to override in order to add that additional logic.
     */
    public function collateUrls(): void
    {
        // Grab any existing Urls so that we can add to it
        $urls = $this->getUrlsToCrawl();

        // Set us to a LIVE stage/reading_mode
        $this->withVersionContext(function() use (&$urls) {
            /** @var SiteTree $owner */
            $owner = $this->getOwner();
            $key = $this->getOwnerKey();

            // We can't do anything if we don't have a key to use
            if ($key === null) {
                return;
            }

            // Create a new container for this key
            if (!array_key_exists($key, $urls)) {
                $urls[$key] = [];
            }

            // Grab the absolute live link without ?stage=Live appended
            $link = $owner->getAbsoluteLiveLink(false);

            // If this record is not published, or we're unable to get a "Live Link" (for whatever reason), then there
            // is nothing more we can do here
            if (!$link) {
                return;
            }

            // Nothing for us to do here, the Link is already being tracked
            if (in_array($link, $urls[$key])) {
                return;
            }

            // Add our base URL to this key
            $urls[$key][] = $link;
        });

        // Update the Urls we have stored for indexing
        $this->setUrlsToCrawl($urls);
    }

    /**
     * Send requests to Swiftype to reindex each of the Urls that we have previously collated
     */
    protected function processCollatedUrls(): void
    {
        // Fetch the Urls that we need to reindex
        $key = $this->getOwnerKey();

        // We can't do anything if we don't have a key to process
        if ($key === null) {
            return;
        }

        $urls = $this->getUrlsToCrawl();

        // There is nothing for us to do here if there are no Urls
        if (count(array_keys($urls)) === 0) {
            return;
        }

        // There are no Urls for this particular key
        if (!array_key_exists($key, $urls)) {
            return;
        }

        // Force the reindexing of each URL we collated
        foreach ($urls[$key] as $url)  {
            $this->forceSwiftypeIndex($url);
        }
    }

    /**
     * @param string $updateUrl
     * @return bool
     */
    protected function forceSwiftypeIndex(string $updateUrl): bool
    {
        // We don't reindex dev environments
        if (Director::isDev()) {
            return true;
        }

        $crawler = SwiftypeCrawler::create();

        return $crawler->send($updateUrl);
    }

    /**
     * @return string
     */
    protected function getOwnerKey(): ?string
    {
        $owner = $this->owner;

        // Can't generate a key if the owner has not yet been written to the DB
        if (!$owner->isInDB()) {
            return null;
        }

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

        return $key;
    }

    /**
     * Sets the version context to Live as that's what crawlers will (normally) see
     *
     * The main function is to suppress the ?stage=Live querystring. LeftAndMain will set the default
     * reading mode to 'DRAFT' when initialising so to counter this we need to re-set the default
     * reading mode back to LIVE
     *
     * @param callable $callback
     */
    private function withVersionContext(callable $callback): void
    {
        Versioned::withVersionedMode(static function() use ($callback) {
            // Grab our current stage and reading mode
            $originalDefaultReadingMode = Versioned::get_default_reading_mode();
            $originalReadingMode = Versioned::get_reading_mode();
            $originalStage = Versioned::get_stage();

            // Set our stage and reading mode to LIVE
            Versioned::set_default_reading_mode('Stage.' . Versioned::LIVE);
            Versioned::set_reading_mode('Stage.' . Versioned::LIVE);
            Versioned::set_stage(Versioned::LIVE);

            // Process whatever callback was provided
            $callback();

            // Set us back to the original stage and reading mode
            if ($originalReadingMode) {
                Versioned::set_default_reading_mode($originalDefaultReadingMode);
                Versioned::set_reading_mode($originalReadingMode);
            }

            if ($originalStage) {
                Versioned::set_stage($originalStage);
            }
        });
    }
}


1			<?php
2
3			namespace Ichaber\SSSwiftype\Extensions;
4
5			use Ichaber\SSSwiftype\Service\SwiftypeCrawler;
6			use SilverStripe\CMS\Model\SiteTree;
7			use SilverStripe\CMS\Model\SiteTreeExtension;
8			use SilverStripe\Control\Director;
9			use SilverStripe\Core\Config\Config;
10			use SilverStripe\Versioned\Versioned;
11
12			/**
13			* Class SwiftypeSiteTreeCrawlerExtension
14			*
15			* @package Ichaber\SSSwiftype\Extensions
16			* @property SiteTree\|$this $owner
17			*/
18			class SwiftypeSiteTreeCrawlerExtension extends SiteTreeExtension
19			{
20			/**
21			* Urls to crawl
22			*
23			* array keyed by getOwnerKey
24			*
25			* @var array
26			*/
27			private $urlsToCrawl = [];
28
29			/**
30			* @param array $urls
31			*/
32			public function setUrlsToCrawl(array $urls) {
33			$this->urlsToCrawl = $urls;
34			}
35
36			/**
37			* @return array
38			*/
39			public function getUrlsToCrawl(): array
40			{
41			return $this->urlsToCrawl;
42			}
43
44			/**
45			* We need to collate Urls before we write, just in case an author has changed the Page's Url Segment. If they
46			* have, then we need to request Swiftype to reindex both the old Url (which should then be marked by Swiftype
47			* as a 404), and the new Url
48			*/
49			public function onBeforeWrite(): void
50			{
51			$this->collateUrls();
52			}
53
54			/**
55			* After a publish has occurred, we can collate and process immediately (no need to split things out like during
56			* an unpublish)
57			*
58			* @param SiteTree\|mixed $original
59			* @return void
60			*/
61			public function onAfterPublish(&$original): void
62			{
63			$this->collateUrls();
64			$this->processCollatedUrls();
65
66			// Check to see if the clearing of cache has been disabled (useful for unit testing, or any other reason you
67			// might have to disable it)
68			$clearCacheDisabled = Config::inst()->get(static::class, 'clear_cache_disabled');
69
70			if ($clearCacheDisabled) {
71			return;
72			}
73
74			// It's important that we clear the cache after we have finished requesting reindex from Swiftype
75			$this->clearCacheSingle();
76			}
77
78			/**
79			* We need to collate the Urls to be purged before we complete the unpublish action (otherwise, the LIVE Urls
80			* will no longer be available, since the page is now unpublished)
81			*/
82			public function onBeforeUnpublish(): void
83			{
84			$this->collateUrls();
85			}
86
87			/**
88			* After the unpublish has completed, we can now request Swiftype to reindex the Urls that we collated
89			*/
90			public function onAfterUnpublish(): void
91			{
92			$this->processCollatedUrls();
93
94			// Check to see if the clearing of cache has been disabled (useful for unit testing, or any other reason you
95			// might have to disable it)
96			$clearCacheDisabled = Config::inst()->get(static::class, 'clear_cache_disabled');
97
98			if ($clearCacheDisabled) {
99			return;
100			}
101
102			// It's important that we clear the cache after we have finished requesting reindex from Swiftype
103			$this->clearCacheSingle();
104			}
105
106			/**
107			* You may need to clear the cache at some point during your particular process
108			*
109			* Reset all Urls for any/all objects that might be in the cache (keeping in mind that Extensions are singleton,
110			* so the UrlsToCache could be accessed via singleton and it could contain Urls for many owner objects)
111			*
112			* We don't use flushCache (which is called from DataObject) because this is called between write and un/publish,
113			* and we need our cache to persist through these states
114			*/
115			public function clearCacheAll(): void
116			{
117			$this->setUrlsToCrawl([]);
118			}
119
120			/**
121			* You may need to clear the cache at some point during your particular process
122			*
123			* Reset only the Urls related to this particular owner object (keeping in mind that Extensions are singleton,
124			* so the UrlsToCache could be accessed via singleton and it could contain Urls for many owner objects)
125			*
126			* We don't use flushCache (which is called from DataObject) because this is called between write and un/publish,
127			* and we need our cache to persist through these states
128			*/
129			public function clearCacheSingle(): void
130			{
131			$urls = $this->getUrlsToCrawl();
132			$key = $this->getOwnerKey();
133
134			// Nothing for us to do here
135			if ($key === null) {
136			return;
137			}
138
139			// Nothing for us to do here
140			if (!array_key_exists($key, $urls)) {
141			return;
142			}
143
144			// Remove this key and it's Urls
145			unset($urls[$key]);
146
147			$this->setUrlsToCrawl($urls);
148			}
149
150			/**
151			* Collate Urls to crawl
152			*
153			* Extensions are singleton, so we use the owner key to make sure that we're only processing Urls directly related
154			* to the desired record.
155			*
156			* You might need to collate more than one URL per Page (maybe you're using Fluent or another translation module).
157			* This is the method you will want to override in order to add that additional logic.
158			*/
159			public function collateUrls(): void
160			{
161			// Grab any existing Urls so that we can add to it
162			$urls = $this->getUrlsToCrawl();
163
164			// Set us to a LIVE stage/reading_mode
165			$this->withVersionContext(function() use (&$urls) {
166			/** @var SiteTree $owner */
167			$owner = $this->getOwner();
168			$key = $this->getOwnerKey();
169
170			// We can't do anything if we don't have a key to use
171			if ($key === null) {
172			return;
173			}
174
175			// Create a new container for this key
176			if (!array_key_exists($key, $urls)) {
177			$urls[$key] = [];
178			}
179
180			// Grab the absolute live link without ?stage=Live appended
181			$link = $owner->getAbsoluteLiveLink(false);
182
183			// If this record is not published, or we're unable to get a "Live Link" (for whatever reason), then there
184			// is nothing more we can do here
185			if (!$link) {
186			return;
187			}
188
189			// Nothing for us to do here, the Link is already being tracked
190			if (in_array($link, $urls[$key])) {
191			return;
192			}
193
194			// Add our base URL to this key
195			$urls[$key][] = $link;
196			});
197
198			// Update the Urls we have stored for indexing
199			$this->setUrlsToCrawl($urls);
200			}
201
202			/**
203			* Send requests to Swiftype to reindex each of the Urls that we have previously collated
204			*/
205			protected function processCollatedUrls(): void
206			{
207			// Fetch the Urls that we need to reindex
208			$key = $this->getOwnerKey();
209
210			// We can't do anything if we don't have a key to process
211			if ($key === null) {
212			return;
213			}
214
215			$urls = $this->getUrlsToCrawl();
216
217			// There is nothing for us to do here if there are no Urls
218			if (count(array_keys($urls)) === 0) {
219			return;
220			}
221
222			// There are no Urls for this particular key
223			if (!array_key_exists($key, $urls)) {
224			return;
225			}
226
227			// Force the reindexing of each URL we collated
228			foreach ($urls[$key] as $url) {
229			$this->forceSwiftypeIndex($url);
230			}
231			}
232
233			/**
234			* @param string $updateUrl
235			* @return bool
236			*/
237			protected function forceSwiftypeIndex(string $updateUrl): bool
238			{
239			// We don't reindex dev environments
240			if (Director::isDev()) {
241			return true;
242			}
243
244			$crawler = SwiftypeCrawler::create();
245
246			return $crawler->send($updateUrl);
247			}
248
249			/**
250			* @return string
251			*/
252			protected function getOwnerKey(): ?string
253			{
254			$owner = $this->owner;
255
256			// Can't generate a key if the owner has not yet been written to the DB
257			if (!$owner->isInDB()) {
258			return null;
259			}
260
261			$key = str_replace('\\', '', $owner->ClassName . $owner->ID);
262
263			return $key;
264			}
265
266			/**
267			* Sets the version context to Live as that's what crawlers will (normally) see
268			*
269			* The main function is to suppress the ?stage=Live querystring. LeftAndMain will set the default
270			* reading mode to 'DRAFT' when initialising so to counter this we need to re-set the default
271			* reading mode back to LIVE
272			*
273			* @param callable $callback
274			*/
275			private function withVersionContext(callable $callback): void
276			{
277			Versioned::withVersionedMode(static function() use ($callback) {
278			// Grab our current stage and reading mode
279			$originalDefaultReadingMode = Versioned::get_default_reading_mode();
280			$originalReadingMode = Versioned::get_reading_mode();
281			$originalStage = Versioned::get_stage();
282
283			// Set our stage and reading mode to LIVE
284			Versioned::set_default_reading_mode('Stage.' . Versioned::LIVE);
285			Versioned::set_reading_mode('Stage.' . Versioned::LIVE);
286			Versioned::set_stage(Versioned::LIVE);
287
288			// Process whatever callback was provided
289			$callback();
290
291			// Set us back to the original stage and reading mode
292			if ($originalReadingMode) {
293			Versioned::set_default_reading_mode($originalDefaultReadingMode);
294			Versioned::set_reading_mode($originalReadingMode);
295			}
296
297			if ($originalStage) {
298			Versioned::set_stage($originalStage);
299			}
300			});
301			}
302			}
303

ichaber / silverstripe-swiftype

Pull Request — master (#16)

SwiftypeSiteTreeCrawlerExtension::setUrlsToCrawl() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like