Crawler::setUserAgent() - Code Metrics - Inspection of "Feature/set user agent" - spatie/crawler - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#246)

unknown

created 2019-07-10 12:08 UTC

Crawler::setUserAgent() A

↳ Parent: Crawler

Complexity

Conditions	1
Paths	1

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	6
rs	10
c	0
b	0
f	0
cc	1
nc	1
nop	1

<?php

namespace Spatie\Crawler;

use Generator;
use Tree\Node\Node;
use GuzzleHttp\Pool;
use GuzzleHttp\Client;
use GuzzleHttp\Psr7\Uri;
use GuzzleHttp\Psr7\Request;
use Spatie\Robots\RobotsTxt;
use GuzzleHttp\RequestOptions;
use Psr\Http\Message\UriInterface;
use Spatie\Browsershot\Browsershot;
use Spatie\Crawler\CrawlQueue\CrawlQueue;
use Spatie\Crawler\Handlers\CrawlRequestFailed;
use Spatie\Crawler\Handlers\CrawlRequestFulfilled;
use Spatie\Crawler\CrawlQueue\CollectionCrawlQueue;
use Spatie\Crawler\Exception\InvalidCrawlRequestHandler;

class Crawler
{
    /** @var \GuzzleHttp\Client */
    protected $client;

    /** @var \Psr\Http\Message\UriInterface */
    protected $baseUrl;

    /** @var \Spatie\Crawler\CrawlObserverCollection */
    protected $crawlObservers;

    /** @var \Spatie\Crawler\CrawlProfile */
    protected $crawlProfile;

    /** @var int */
    protected $concurrency;

    /** @var \Spatie\Crawler\CrawlQueue\CrawlQueue */
    protected $crawlQueue;

    /** @var int */
    protected $crawledUrlCount = 0;

    /** @var int|null */
    protected $maximumCrawlCount = null;

    /** @var int */
    protected $maximumResponseSize = 1024 * 1024 * 2;

    /** @var int|null */
    protected $maximumDepth = null;

    /** @var bool */
    protected $respectRobots = true;

    /** @var \Tree\Node\Node */
    protected $depthTree;

    /** @var bool */
    protected $executeJavaScript = false;

    /** @var Browsershot */
    protected $browsershot = null;

    /** @var \Spatie\Robots\RobotsTxt */
    protected $robotsTxt = null;

    /** @var string */
    protected $crawlRequestFulfilledClass;

    /** @var string */
    protected $crawlRequestFailedClass;

    /** @var int */
    protected $delayBetweenRequests = 0;

    /** @var string */
    protected $userAgent = '*';

    /** @var   */
    protected static $defaultClientOptions = [
        RequestOptions::COOKIES => true,
        RequestOptions::CONNECT_TIMEOUT => 10,
        RequestOptions::TIMEOUT => 10,
        RequestOptions::ALLOW_REDIRECTS => false,
    ];

    public static function create(array $clientOptions = []): Crawler
    {
        $clientOptions = (count($clientOptions))
            ? $clientOptions
            : static::$defaultClientOptions;

        $client = new Client($clientOptions);

        return new static($client);
    }

    public function __construct(Client $client, int $concurrency = 10)
    {
        $this->client = $client;

        $this->concurrency = $concurrency;

        $this->crawlProfile = new CrawlAllUrls();

        $this->crawlQueue = new CollectionCrawlQueue();

        $this->crawlObservers = new CrawlObserverCollection();

        $this->crawlRequestFulfilledClass = CrawlRequestFulfilled::class;

        $this->crawlRequestFailedClass = CrawlRequestFailed::class;
    }

    public function setConcurrency(int $concurrency): Crawler
    {
        $this->concurrency = $concurrency;

        return $this;
    }

    public function setMaximumResponseSize(int $maximumResponseSizeInBytes): Crawler
    {
        $this->maximumResponseSize = $maximumResponseSizeInBytes;

        return $this;
    }

    public function getMaximumResponseSize(): ?int
    {
        return $this->maximumResponseSize;
    }

    public function setMaximumCrawlCount(int $maximumCrawlCount): Crawler
    {
        $this->maximumCrawlCount = $maximumCrawlCount;

        return $this;
    }

    public function getMaximumCrawlCount(): ?int
    {
        return $this->maximumCrawlCount;
    }

    public function getCrawlerUrlCount(): int
    {
        return $this->crawledUrlCount;
    }

    public function setMaximumDepth(int $maximumDepth): Crawler
    {
        $this->maximumDepth = $maximumDepth;

        return $this;
    }

    public function getMaximumDepth(): ?int
    {
        return $this->maximumDepth;
    }

    /**
     * @param int $delay The delay in milliseconds.
     *
     * @return Crawler
     */
    public function setDelayBetweenRequests(int $delay): Crawler
    {
        $this->delayBetweenRequests = ($delay * 1000);

        return $this;
    }

    /**
     * @return int The delay in milliseconds.
     */
    public function getDelayBetweenRequests(): int
    {
        return $this->delayBetweenRequests;
    }

    public function ignoreRobots(): Crawler
    {
        $this->respectRobots = false;

        return $this;
    }

    public function respectRobots(): Crawler
    {
        $this->respectRobots = true;

        return $this;
    }

    public function mustRespectRobots(): bool
    {
        return $this->respectRobots;
    }

    public function getRobotsTxt(): RobotsTxt
    {
        return $this->robotsTxt;
    }

    public function setCrawlQueue(CrawlQueue $crawlQueue): Crawler
    {
        $this->crawlQueue = $crawlQueue;

        return $this;
    }

    public function getCrawlQueue(): CrawlQueue
    {
        return $this->crawlQueue;
    }

    public function executeJavaScript(): Crawler
    {
        $this->executeJavaScript = true;

        return $this;
    }

    public function doNotExecuteJavaScript(): Crawler
    {
        $this->executeJavaScript = false;

        return $this;
    }

    public function mayExecuteJavascript(): bool
    {
        return $this->executeJavaScript;
    }

    /**
     * @param \Spatie\Crawler\CrawlObserver|array[\Spatie\Crawler\CrawlObserver] $crawlObservers

     *
     * @return $this
     */
    public function setCrawlObserver($crawlObservers): Crawler
    {
        if (! is_array($crawlObservers)) {
            $crawlObservers = [$crawlObservers];
        }

        return $this->setCrawlObservers($crawlObservers);
    }

    public function setCrawlObservers(array $crawlObservers): Crawler
    {
        $this->crawlObservers = new CrawlObserverCollection($crawlObservers);

        return $this;
    }

    public function addCrawlObserver(CrawlObserver $crawlObserver): Crawler
    {
        $this->crawlObservers->addObserver($crawlObserver);

        return $this;
    }

    public function getCrawlObservers(): CrawlObserverCollection
    {
        return $this->crawlObservers;
    }

    public function setCrawlProfile(CrawlProfile $crawlProfile): Crawler
    {
        $this->crawlProfile = $crawlProfile;

        return $this;
    }

    public function getCrawlProfile(): CrawlProfile
    {
        return $this->crawlProfile;
    }

    public function setCrawlFulfilledHandlerClass(string $crawlRequestFulfilledClass): Crawler
    {
        $baseClass = CrawlRequestFulfilled::class;

        if (! is_subclass_of($crawlRequestFulfilledClass, $baseClass)) {
            throw InvalidCrawlRequestHandler::doesNotExtendBaseClass($crawlRequestFulfilledClass, $baseClass);
        }

        $this->crawlRequestFulfilledClass = $crawlRequestFulfilledClass;

        return $this;
    }

    public function setCrawlFailedHandlerClass(string $crawlRequestFailedClass): Crawler
    {
        $baseClass = CrawlRequestFailed::class;

        if (! is_subclass_of($crawlRequestFailedClass, $baseClass)) {
            throw InvalidCrawlRequestHandler::doesNotExtendBaseClass($crawlRequestFailedClass, $baseClass);
        }

        $this->crawlRequestFailedClass = $crawlRequestFailedClass;

        return $this;
    }

    public function setBrowsershot(Browsershot $browsershot)
    {
        $this->browsershot = $browsershot;

        return $this;
    }

    public function setUserAgent(string $userAgent): Crawler
    {
        $this->userAgent = strtolower($userAgent);

        return $this;
    }

    public function getUserAgent(): string
    {
        return $this->userAgent;
    }

    public function getBrowsershot(): Browsershot
    {
        if (! $this->browsershot) {
            $this->browsershot = new Browsershot();
        }

        return $this->browsershot;
    }

    public function getBaseUrl(): UriInterface
    {
        return $this->baseUrl;
    }

    /**
     * @param \Psr\Http\Message\UriInterface|string $baseUrl
     */
    public function startCrawling($baseUrl)
    {
        if (! $baseUrl instanceof UriInterface) {
            $baseUrl = new Uri($baseUrl);
        }

        if ($baseUrl->getScheme() === '') {
            $baseUrl = $baseUrl->withScheme('http');
        }

        if ($baseUrl->getPath() === '') {
            $baseUrl = $baseUrl->withPath('/');
        }

        $this->baseUrl = $baseUrl;

        $crawlUrl = CrawlUrl::create($this->baseUrl);

        $this->robotsTxt = $this->createRobotsTxt($crawlUrl->url);

        if ($this->robotsTxt->allows((string) $crawlUrl->url, (string) $this->userAgent) ||
            ! $this->respectRobots
        ) {
            $this->addToCrawlQueue($crawlUrl);
        }

        $this->depthTree = new Node((string) $this->baseUrl);

        $this->startCrawlingQueue();

        foreach ($this->crawlObservers as $crawlObserver) {
            $crawlObserver->finishedCrawling();
        }
    }

    public function addToDepthTree(UriInterface $url, UriInterface $parentUrl, Node $node = null): ?Node
    {
        if (is_null($this->maximumDepth)) {
            return new Node((string) $url);
        }

        $node = $node ?? $this->depthTree;

        $returnNode = null;

        if ($node->getValue() === (string) $parentUrl) {
            $newNode = new Node((string) $url);

            $node->addChild($newNode);

            return $newNode;
        }

        foreach ($node->getChildren() as $currentNode) {
            $returnNode = $this->addToDepthTree($url, $parentUrl, $currentNode);

            if (! is_null($returnNode)) {
                break;
            }
        }

        return $returnNode;
    }

    protected function startCrawlingQueue()
    {
        while ($this->crawlQueue->hasPendingUrls()) {
            $pool = new Pool($this->client, $this->getCrawlRequests(), [
                'concurrency' => $this->concurrency,
                'options' => $this->client->getConfig(),
                'fulfilled' => new $this->crawlRequestFulfilledClass($this),
                'rejected' => new $this->crawlRequestFailedClass($this),
            ]);

            $promise = $pool->promise();

            $promise->wait();
        }
    }

    /**
     * @deprecated This function will be removed in the next major version
     */
    public function endsWith($haystack, $needle)
    {
        return strrpos($haystack, $needle) + strlen($needle) ===
            strlen($haystack);
    }

    protected function createRobotsTxt(UriInterface $uri): RobotsTxt
    {
        return RobotsTxt::create($uri->withPath('/robots.txt'));
    }

    protected function getCrawlRequests(): Generator
    {
        while ($crawlUrl = $this->crawlQueue->getFirstPendingUrl()) {
            if (! $this->crawlProfile->shouldCrawl($crawlUrl->url)) {
                $this->crawlQueue->markAsProcessed($crawlUrl);
                continue;
            }

            if ($this->crawlQueue->hasAlreadyBeenProcessed($crawlUrl)) {
                continue;
            }

            foreach ($this->crawlObservers as $crawlObserver) {
                $crawlObserver->willCrawl($crawlUrl->url);
            }

            $this->crawlQueue->markAsProcessed($crawlUrl);

            yield $crawlUrl->getId() => new Request('GET', $crawlUrl->url);
        }
    }

    public function addToCrawlQueue(CrawlUrl $crawlUrl): Crawler
    {
        if (! $this->getCrawlProfile()->shouldCrawl($crawlUrl->url)) {
            return $this;
        }

        if ($this->getCrawlQueue()->has($crawlUrl->url)) {
            return $this;
        }

        $this->crawledUrlCount++;

        $this->crawlQueue->add($crawlUrl);

        return $this;
    }

    public function maximumCrawlCountReached(): bool
    {
        $maximumCrawlCount = $this->getMaximumCrawlCount();

        if (is_null($maximumCrawlCount)) {
            return false;
        }

        return $this->getCrawlerUrlCount() >= $maximumCrawlCount;
    }
}


1			<?php
2
3			namespace Spatie\Crawler;
4
5			use Generator;
6			use Tree\Node\Node;
7			use GuzzleHttp\Pool;
8			use GuzzleHttp\Client;
9			use GuzzleHttp\Psr7\Uri;
10			use GuzzleHttp\Psr7\Request;
11			use Spatie\Robots\RobotsTxt;
12			use GuzzleHttp\RequestOptions;
13			use Psr\Http\Message\UriInterface;
14			use Spatie\Browsershot\Browsershot;
15			use Spatie\Crawler\CrawlQueue\CrawlQueue;
16			use Spatie\Crawler\Handlers\CrawlRequestFailed;
17			use Spatie\Crawler\Handlers\CrawlRequestFulfilled;
18			use Spatie\Crawler\CrawlQueue\CollectionCrawlQueue;
19			use Spatie\Crawler\Exception\InvalidCrawlRequestHandler;
20
21			class Crawler
22			{
23			/** @var \GuzzleHttp\Client */
24			protected $client;
25
26			/** @var \Psr\Http\Message\UriInterface */
27			protected $baseUrl;
28
29			/** @var \Spatie\Crawler\CrawlObserverCollection */
30			protected $crawlObservers;
31
32			/** @var \Spatie\Crawler\CrawlProfile */
33			protected $crawlProfile;
34
35			/** @var int */
36			protected $concurrency;
37
38			/** @var \Spatie\Crawler\CrawlQueue\CrawlQueue */
39			protected $crawlQueue;
40
41			/** @var int */
42			protected $crawledUrlCount = 0;
43
44			/** @var int\|null */
45			protected $maximumCrawlCount = null;
46
47			/** @var int */
48			protected $maximumResponseSize = 1024 * 1024 * 2;
49
50			/** @var int\|null */
51			protected $maximumDepth = null;
52
53			/** @var bool */
54			protected $respectRobots = true;
55
56			/** @var \Tree\Node\Node */
57			protected $depthTree;
58
59			/** @var bool */
60			protected $executeJavaScript = false;
61
62			/** @var Browsershot */
63			protected $browsershot = null;
64
65			/** @var \Spatie\Robots\RobotsTxt */
66			protected $robotsTxt = null;
67
68			/** @var string */
69			protected $crawlRequestFulfilledClass;
70
71			/** @var string */
72			protected $crawlRequestFailedClass;
73
74			/** @var int */
75			protected $delayBetweenRequests = 0;
76
77			/** @var string */
78			protected $userAgent = '*';
79
80			/** @var */
81			protected static $defaultClientOptions = [
82			RequestOptions::COOKIES => true,
83			RequestOptions::CONNECT_TIMEOUT => 10,
84			RequestOptions::TIMEOUT => 10,
85			RequestOptions::ALLOW_REDIRECTS => false,
86			];
87
88			public static function create(array $clientOptions = []): Crawler
89			{
90			$clientOptions = (count($clientOptions))
91			? $clientOptions
92			: static::$defaultClientOptions;
93
94			$client = new Client($clientOptions);
95
96			return new static($client);
97			}
98
99			public function __construct(Client $client, int $concurrency = 10)
100			{
101			$this->client = $client;
102
103			$this->concurrency = $concurrency;
104
105			$this->crawlProfile = new CrawlAllUrls();
106
107			$this->crawlQueue = new CollectionCrawlQueue();
108
109			$this->crawlObservers = new CrawlObserverCollection();
110
111			$this->crawlRequestFulfilledClass = CrawlRequestFulfilled::class;
112
113			$this->crawlRequestFailedClass = CrawlRequestFailed::class;
114			}
115
116			public function setConcurrency(int $concurrency): Crawler
117			{
118			$this->concurrency = $concurrency;
119
120			return $this;
121			}
122
123			public function setMaximumResponseSize(int $maximumResponseSizeInBytes): Crawler
124			{
125			$this->maximumResponseSize = $maximumResponseSizeInBytes;
126
127			return $this;
128			}
129
130			public function getMaximumResponseSize(): ?int
131			{
132			return $this->maximumResponseSize;
133			}
134
135			public function setMaximumCrawlCount(int $maximumCrawlCount): Crawler
136			{
137			$this->maximumCrawlCount = $maximumCrawlCount;
138
139			return $this;
140			}
141
142			public function getMaximumCrawlCount(): ?int
143			{
144			return $this->maximumCrawlCount;
145			}
146
147			public function getCrawlerUrlCount(): int
148			{
149			return $this->crawledUrlCount;
150			}
151
152			public function setMaximumDepth(int $maximumDepth): Crawler
153			{
154			$this->maximumDepth = $maximumDepth;
155
156			return $this;
157			}
158
159			public function getMaximumDepth(): ?int
160			{
161			return $this->maximumDepth;
162			}
163
164			/**
165			* @param int $delay The delay in milliseconds.
166			*
167			* @return Crawler
168			*/
169			public function setDelayBetweenRequests(int $delay): Crawler
170			{
171			$this->delayBetweenRequests = ($delay * 1000);
172
173			return $this;
174			}
175
176			/**
177			* @return int The delay in milliseconds.
178			*/
179			public function getDelayBetweenRequests(): int
180			{
181			return $this->delayBetweenRequests;
182			}
183
184			public function ignoreRobots(): Crawler
185			{
186			$this->respectRobots = false;
187
188			return $this;
189			}
190
191			public function respectRobots(): Crawler
192			{
193			$this->respectRobots = true;
194
195			return $this;
196			}
197
198			public function mustRespectRobots(): bool
199			{
200			return $this->respectRobots;
201			}
202
203			public function getRobotsTxt(): RobotsTxt
204			{
205			return $this->robotsTxt;
206			}
207
208			public function setCrawlQueue(CrawlQueue $crawlQueue): Crawler
209			{
210			$this->crawlQueue = $crawlQueue;
211
212			return $this;
213			}
214
215			public function getCrawlQueue(): CrawlQueue
216			{
217			return $this->crawlQueue;
218			}
219
220			public function executeJavaScript(): Crawler
221			{
222			$this->executeJavaScript = true;
223
224			return $this;
225			}
226
227			public function doNotExecuteJavaScript(): Crawler
228			{
229			$this->executeJavaScript = false;
230
231			return $this;
232			}
233
234			public function mayExecuteJavascript(): bool
235			{
236			return $this->executeJavaScript;
237			}
238
239			/**
240			* @param \Spatie\Crawler\CrawlObserver\|array[\Spatie\Crawler\CrawlObserver] $crawlObservers
			0 ignored issues – show Documentation introduced 2018-01-25 20:52 UTC by Report Bug Copy Issue Report The doc-type `\Spatie\Crawler\CrawlObs...\Crawler\CrawlObserver]` could not be parsed: Expected "]" at position 4, but found "\Spatie\Crawler\CrawlObserver". (view supported doc-types) This check marks PHPDoc comments that could not be parsed by our parser. To see which comment annotations we can parse, please refer to our documentation on supported doc-types. Loading history...
241			*
242			* @return $this
243			*/
244			public function setCrawlObserver($crawlObservers): Crawler
245			{
246			if (! is_array($crawlObservers)) {
247			$crawlObservers = [$crawlObservers];
248			}
249
250			return $this->setCrawlObservers($crawlObservers);
251			}
252
253			public function setCrawlObservers(array $crawlObservers): Crawler
254			{
255			$this->crawlObservers = new CrawlObserverCollection($crawlObservers);
256
257			return $this;
258			}
259
260			public function addCrawlObserver(CrawlObserver $crawlObserver): Crawler
261			{
262			$this->crawlObservers->addObserver($crawlObserver);
263
264			return $this;
265			}
266
267			public function getCrawlObservers(): CrawlObserverCollection
268			{
269			return $this->crawlObservers;
270			}
271
272			public function setCrawlProfile(CrawlProfile $crawlProfile): Crawler
273			{
274			$this->crawlProfile = $crawlProfile;
275
276			return $this;
277			}
278
279			public function getCrawlProfile(): CrawlProfile
280			{
281			return $this->crawlProfile;
282			}
283
284			public function setCrawlFulfilledHandlerClass(string $crawlRequestFulfilledClass): Crawler
285			{
286			$baseClass = CrawlRequestFulfilled::class;
287
288			if (! is_subclass_of($crawlRequestFulfilledClass, $baseClass)) {
289			throw InvalidCrawlRequestHandler::doesNotExtendBaseClass($crawlRequestFulfilledClass, $baseClass);
290			}
291
292			$this->crawlRequestFulfilledClass = $crawlRequestFulfilledClass;
293
294			return $this;
295			}
296
297			public function setCrawlFailedHandlerClass(string $crawlRequestFailedClass): Crawler
298			{
299			$baseClass = CrawlRequestFailed::class;
300
301			if (! is_subclass_of($crawlRequestFailedClass, $baseClass)) {
302			throw InvalidCrawlRequestHandler::doesNotExtendBaseClass($crawlRequestFailedClass, $baseClass);
303			}
304
305			$this->crawlRequestFailedClass = $crawlRequestFailedClass;
306
307			return $this;
308			}
309
310			public function setBrowsershot(Browsershot $browsershot)
311			{
312			$this->browsershot = $browsershot;
313
314			return $this;
315			}
316
317			public function setUserAgent(string $userAgent): Crawler
318			{
319			$this->userAgent = strtolower($userAgent);
320
321			return $this;
322			}
323
324			public function getUserAgent(): string
325			{
326			return $this->userAgent;
327			}
328
329			public function getBrowsershot(): Browsershot
330			{
331			if (! $this->browsershot) {
332			$this->browsershot = new Browsershot();
333			}
334
335			return $this->browsershot;
336			}
337
338			public function getBaseUrl(): UriInterface
339			{
340			return $this->baseUrl;
341			}
342
343			/**
344			* @param \Psr\Http\Message\UriInterface\|string $baseUrl
345			*/
346			public function startCrawling($baseUrl)
347			{
348			if (! $baseUrl instanceof UriInterface) {
349			$baseUrl = new Uri($baseUrl);
350			}
351
352			if ($baseUrl->getScheme() === '') {
353			$baseUrl = $baseUrl->withScheme('http');
354			}
355
356			if ($baseUrl->getPath() === '') {
357			$baseUrl = $baseUrl->withPath('/');
358			}
359
360			$this->baseUrl = $baseUrl;
361
362			$crawlUrl = CrawlUrl::create($this->baseUrl);
363
364			$this->robotsTxt = $this->createRobotsTxt($crawlUrl->url);
365
366			if ($this->robotsTxt->allows((string) $crawlUrl->url, (string) $this->userAgent) \|\|
367			! $this->respectRobots
368			) {
369			$this->addToCrawlQueue($crawlUrl);
370			}
371
372			$this->depthTree = new Node((string) $this->baseUrl);
373
374			$this->startCrawlingQueue();
375
376			foreach ($this->crawlObservers as $crawlObserver) {
377			$crawlObserver->finishedCrawling();
378			}
379			}
380
381			public function addToDepthTree(UriInterface $url, UriInterface $parentUrl, Node $node = null): ?Node
382			{
383			if (is_null($this->maximumDepth)) {
384			return new Node((string) $url);
385			}
386
387			$node = $node ?? $this->depthTree;
388
389			$returnNode = null;
390
391			if ($node->getValue() === (string) $parentUrl) {
392			$newNode = new Node((string) $url);
393
394			$node->addChild($newNode);
395
396			return $newNode;
397			}
398
399			foreach ($node->getChildren() as $currentNode) {
400			$returnNode = $this->addToDepthTree($url, $parentUrl, $currentNode);
401
402			if (! is_null($returnNode)) {
403			break;
404			}
405			}
406
407			return $returnNode;
408			}
409
410			protected function startCrawlingQueue()
411			{
412			while ($this->crawlQueue->hasPendingUrls()) {
413			$pool = new Pool($this->client, $this->getCrawlRequests(), [
414			'concurrency' => $this->concurrency,
415			'options' => $this->client->getConfig(),
416			'fulfilled' => new $this->crawlRequestFulfilledClass($this),
417			'rejected' => new $this->crawlRequestFailedClass($this),
418			]);
419
420			$promise = $pool->promise();
421
422			$promise->wait();
423			}
424			}
425
426			/**
427			* @deprecated This function will be removed in the next major version
428			*/
429			public function endsWith($haystack, $needle)
430			{
431			return strrpos($haystack, $needle) + strlen($needle) ===
432			strlen($haystack);
433			}
434
435			protected function createRobotsTxt(UriInterface $uri): RobotsTxt
436			{
437			return RobotsTxt::create($uri->withPath('/robots.txt'));
438			}
439
440			protected function getCrawlRequests(): Generator
441			{
442			while ($crawlUrl = $this->crawlQueue->getFirstPendingUrl()) {
443			if (! $this->crawlProfile->shouldCrawl($crawlUrl->url)) {
444			$this->crawlQueue->markAsProcessed($crawlUrl);
445			continue;
446			}
447
448			if ($this->crawlQueue->hasAlreadyBeenProcessed($crawlUrl)) {
449			continue;
450			}
451
452			foreach ($this->crawlObservers as $crawlObserver) {
453			$crawlObserver->willCrawl($crawlUrl->url);
454			}
455
456			$this->crawlQueue->markAsProcessed($crawlUrl);
457
458			yield $crawlUrl->getId() => new Request('GET', $crawlUrl->url);
459			}
460			}
461
462			public function addToCrawlQueue(CrawlUrl $crawlUrl): Crawler
463			{
464			if (! $this->getCrawlProfile()->shouldCrawl($crawlUrl->url)) {
465			return $this;
466			}
467
468			if ($this->getCrawlQueue()->has($crawlUrl->url)) {
469			return $this;
470			}
471
472			$this->crawledUrlCount++;
473
474			$this->crawlQueue->add($crawlUrl);
475
476			return $this;
477			}
478
479			public function maximumCrawlCountReached(): bool
480			{
481			$maximumCrawlCount = $this->getMaximumCrawlCount();
482
483			if (is_null($maximumCrawlCount)) {
484			return false;
485			}
486
487			return $this->getCrawlerUrlCount() >= $maximumCrawlCount;
488			}
489			}
490

spatie / crawler

Pull Request — master (#246)

Crawler::setUserAgent() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like