Crawler::extractAllLinks() - Code Metrics - Inspection of "Implementing depth" - spatie/crawler - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#65)

unknown

created 2017-09-26 16:10 UTC

Crawler::extractAllLinks() A

↳ Parent: Crawler

Complexity

Conditions	2
Paths	2

Size

Total Lines	13
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
c	0
b	0
f	0
dl	0
loc	13
rs	9.4285
cc	2
eloc	7
nc	2
nop	2

<?php

namespace Spatie\Crawler;

use Generator;
use GuzzleHttp\Pool;
use GuzzleHttp\Client;
use GuzzleHttp\Psr7\Request;
use GuzzleHttp\RequestOptions;
use Illuminate\Support\Collection;
use Spatie\Browsershot\Browsershot;
use Symfony\Component\DomCrawler\Link;
use Psr\Http\Message\ResponseInterface;
use GuzzleHttp\Exception\RequestException;
use Symfony\Component\DomCrawler\Crawler as DomCrawler;
use Tree\Node\Node;

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

    /** @var \Spatie\Crawler\Url */
    protected $baseUrl;

    /** @var \Spatie\Crawler\CrawlObserver */
    protected $crawlObserver;

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

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

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

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

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

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

    /** @var string|null */
    protected $pathToChromeBinary = null;

    /**
     * @param array $clientOptions
     *
     * @return static
     */
    public static function create(array $clientOptions = [])
    {
        $hasClientOpts = (bool) count($clientOptions);
        $client = new Client($hasClientOpts ? $clientOptions : [
                RequestOptions::COOKIES => true,
                RequestOptions::CONNECT_TIMEOUT => 10,
                RequestOptions::TIMEOUT => 10,
                RequestOptions::ALLOW_REDIRECTS => false,
            ]);

        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 CrawlQueue();
    }

    /**
     * @param int $concurrency
     *
     * @return $this
     */
    public function setConcurrency(int $concurrency)
    {
        $this->concurrency = $concurrency;

        return $this;
    }

	/**
	 * @param int $maximumDepth
	 *
	 * @return $this
	 */
	public function setMaximumDepth(int $maximumDepth) {

		$this->maximumDepth = $maximumDepth;

		return $this;
	}

    /**
     * @return $this
     */
    public function executeJavaScript($pathToChromeBinary = null)
    {
        $this->executeJavaScript = true;
$answer = 42;

$correct = false;

$correct = (bool) $answer;

        $this->pathToChromeBinary = $pathToChromeBinary;

        return $this;
    }

    /**
     * @return $this
     */
    public function doNotExecuteJavaScript()
    {
        $this->executeJavaScript = false;

        return $this;
    }

    /**
     * @param \Spatie\Crawler\CrawlObserver $crawlObserver
     *
     * @return $this
     */
    public function setCrawlObserver(CrawlObserver $crawlObserver)
    {
        $this->crawlObserver = $crawlObserver;

        return $this;
    }

    /**
     * @param \Spatie\Crawler\CrawlProfile $crawlProfile
     *
     * @return $this
     */
    public function setCrawlProfile(CrawlProfile $crawlProfile)
    {
        $this->crawlProfile = $crawlProfile;

        return $this;
    }

    /**
     * @param \Spatie\Crawler\Url|string $baseUrl
     */
    public function startCrawling($baseUrl)
    {
        if (! $baseUrl instanceof Url) {
            $baseUrl = Url::create($baseUrl);
        }

        $this->baseUrl = $baseUrl;

        $crawlUrl = CrawlUrl::create($baseUrl);

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

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

        $this->startCrawlingQueue();

        $this->crawlObserver->finishedCrawling();
    }

    protected function startCrawlingQueue()
    {
        while ($this->crawlQueue->hasPendingUrls()) {
            $pool = new Pool($this->client, $this->getCrawlRequests(), [
                'concurrency' => $this->concurrency,
                'options' => $this->client->getConfig(),
                'fulfilled' => function (ResponseInterface $response, int $index) {
                    $this->handleResponse($response, $index);

                    $crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($index);

                    if ($crawlUrl->url->host !== $this->baseUrl->host) {
                        return;
                    }

                    $this->addAllLinksToCrawlQueue(
                        (string) $response->getBody(),
                        $crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($index)->url
                    );
                },
                'rejected' => function (RequestException $exception, int $index) {
                    $this->handleResponse($exception->getResponse(), $index);
                },
            ]);

            $promise = $pool->promise();
            $promise->wait();

            $this->crawlQueue->removeProcessedUrlsFromPending();
        }
    }

    /**
     * @param ResponseInterface|null $response
     * @param int $index
     */
    protected function handleResponse($response, int $index)
    {
        $crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($index);

        $this->crawlObserver->hasBeenCrawled($crawlUrl->url, $response, $crawlUrl->foundOnUrl);
    }

    protected function getCrawlRequests(): Generator
    {
        $i = 0;
        while ($crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($i)) {
            if (! $this->crawlProfile->shouldCrawl($crawlUrl->url)) {
                $i++;
                continue;
            }

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

            $this->crawlObserver->willCrawl($crawlUrl->url);

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

            yield new Request('GET', (string) $crawlUrl->url);
            $i++;
        }
    }

    protected function addAllLinksToCrawlQueue(string $html, Url $foundOnUrl)
    {
        $allLinks = $this->extractAllLinks($html, $foundOnUrl);

        collect($allLinks)
            ->filter(function (Url $url) {
                return $url->hasCrawlableScheme();
            })
            ->map(function (Url $url) use ($foundOnUrl) {
                return $this->normalizeUrl($url);
            })
            ->filter(function (Url $url) {
                return $this->crawlProfile->shouldCrawl($url);
            })
            ->reject(function ($url) {
                return $this->crawlQueue->has($url);
            })
            ->each(function (Url $url) use ($foundOnUrl) {
				$node = $this->addToLinkTree($this->linkTree, (string)$url, $foundOnUrl);

				if(($this->maximumDepth == 0) || ($node->getDepth() <= $this->maximumDepth)) {
					$this->crawlQueue->add(
						CrawlUrl::create($url, $foundOnUrl)
					);
				}
            });
    }

    protected function extractAllLinks(string $html, Url $foundOnUrl): Collection
    {
        if ($this->executeJavaScript) {
            $html = $this->getBodyAfterExecutingJavaScript($foundOnUrl);
        }

        $domCrawler = new DomCrawler($html, $foundOnUrl);

        return collect($domCrawler->filterXpath('//a')->links())
            ->map(function (Link $link) {
                return Url::create($link->getUri());
            });
    }

    protected function normalizeUrl(Url $url): Url
    {
        return $url->removeFragment();
    }

	/**
	 * @param $node \Tree\Node\Node
	 * @param $url string
	 * @param $parentUrl string
	 */
	protected function addToLinkTree(Node $node, string $url, string $parentUrl) {

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

			return $newNode;
		}
		foreach($node->getChildren() as $currentNode) {
			$returnNode = $this->addToLinkTree($currentNode, $url, $parentUrl);
			if($returnNode !== null) {
				break;
			}
		}

		return $returnNode;
	}

    protected function getBodyAfterExecutingJavaScript(Url $foundOnUrl): string
    {
        $browsershot = Browsershot::url((string) $foundOnUrl);

        if ($this->pathToChromeBinary) {
''   == false // true
''   == null  // true
'ab' == false // false
'ab' == null  // false

// It is often better to use strict comparison
'' === false // false
'' === null  // false
            $browsershot->setChromePath($this->pathToChromeBinary);
        }

        $html = $browsershot->bodyHtml();

        return html_entity_decode($html);
    }

}


1			<?php
2
3			namespace Spatie\Crawler;
4
5			use Generator;
6			use GuzzleHttp\Pool;
7			use GuzzleHttp\Client;
8			use GuzzleHttp\Psr7\Request;
9			use GuzzleHttp\RequestOptions;
10			use Illuminate\Support\Collection;
11			use Spatie\Browsershot\Browsershot;
12			use Symfony\Component\DomCrawler\Link;
13			use Psr\Http\Message\ResponseInterface;
14			use GuzzleHttp\Exception\RequestException;
15			use Symfony\Component\DomCrawler\Crawler as DomCrawler;
16			use Tree\Node\Node;
17
18			class Crawler
19			{
20			/** @var \GuzzleHttp\Client */
21			protected $client;
22
23			/** @var \Spatie\Crawler\Url */
24			protected $baseUrl;
25
26			/** @var \Spatie\Crawler\CrawlObserver */
27			protected $crawlObserver;
28
29			/** @var \Spatie\Crawler\CrawlProfile */
30			protected $crawlProfile;
31
32			/** @var int */
33			protected $concurrency;
34
35			/** @var \Spatie\Crawler\CrawlQueue */
36			protected $crawlQueue;
37
38			/** @var int */
39			protected $maximumDepth = 0;
40
41			/** @var \Tree\Node\Node */
42			protected $linkTree;
43
44			/** @var false */
45			protected $executeJavaScript = false;
46
47			/** @var string\|null */
48			protected $pathToChromeBinary = null;
49
50			/**
51			* @param array $clientOptions
52			*
53			* @return static
54			*/
55			public static function create(array $clientOptions = [])
56			{
57			$hasClientOpts = (bool) count($clientOptions);
58			$client = new Client($hasClientOpts ? $clientOptions : [
59			RequestOptions::COOKIES => true,
60			RequestOptions::CONNECT_TIMEOUT => 10,
61			RequestOptions::TIMEOUT => 10,
62			RequestOptions::ALLOW_REDIRECTS => false,
63			]);
64
65			return new static($client);
66			}
67
68			public function __construct(Client $client, int $concurrency = 10)
69			{
70			$this->client = $client;
71
72			$this->concurrency = $concurrency;
73
74			$this->crawlProfile = new CrawlAllUrls();
75
76			$this->crawlQueue = new CrawlQueue();
77			}
78
79			/**
80			* @param int $concurrency
81			*
82			* @return $this
83			*/
84			public function setConcurrency(int $concurrency)
85			{
86			$this->concurrency = $concurrency;
87
88			return $this;
89			}
90
91			/**
92			* @param int $maximumDepth
93			*
94			* @return $this
95			*/
96			public function setMaximumDepth(int $maximumDepth) {
97
98			$this->maximumDepth = $maximumDepth;
99
100			return $this;
101			}
102
103			/**
104			* @return $this
105			*/
106			public function executeJavaScript($pathToChromeBinary = null)
107			{
108			$this->executeJavaScript = true;
			0 ignored issues – show Documentation Bug introduced 2017-09-21 20:55 UTC by Report Bug Copy Issue Report The property `$executeJavaScript` was declared of type `false`, but `true` is of type `boolean`. Maybe add a type cast? This check looks for assignments to scalar types that may be of the wrong type. To ensure the code behaves as expected, it may be a good idea to add an explicit type cast. $answer = 42; $correct = false; $correct = (bool) $answer; Loading history...
109
110			$this->pathToChromeBinary = $pathToChromeBinary;
111
112			return $this;
113			}
114
115			/**
116			* @return $this
117			*/
118			public function doNotExecuteJavaScript()
119			{
120			$this->executeJavaScript = false;
121
122			return $this;
123			}
124
125			/**
126			* @param \Spatie\Crawler\CrawlObserver $crawlObserver
127			*
128			* @return $this
129			*/
130			public function setCrawlObserver(CrawlObserver $crawlObserver)
131			{
132			$this->crawlObserver = $crawlObserver;
133
134			return $this;
135			}
136
137			/**
138			* @param \Spatie\Crawler\CrawlProfile $crawlProfile
139			*
140			* @return $this
141			*/
142			public function setCrawlProfile(CrawlProfile $crawlProfile)
143			{
144			$this->crawlProfile = $crawlProfile;
145
146			return $this;
147			}
148
149			/**
150			* @param \Spatie\Crawler\Url\|string $baseUrl
151			*/
152			public function startCrawling($baseUrl)
153			{
154			if (! $baseUrl instanceof Url) {
155			$baseUrl = Url::create($baseUrl);
156			}
157
158			$this->baseUrl = $baseUrl;
159
160			$crawlUrl = CrawlUrl::create($baseUrl);
161
162			$this->crawlQueue->add($crawlUrl);
163
164			$this->linkTree = new Node((string)$this->baseUrl);
165
166			$this->startCrawlingQueue();
167
168			$this->crawlObserver->finishedCrawling();
169			}
170
171			protected function startCrawlingQueue()
172			{
173			while ($this->crawlQueue->hasPendingUrls()) {
174			$pool = new Pool($this->client, $this->getCrawlRequests(), [
175			'concurrency' => $this->concurrency,
176			'options' => $this->client->getConfig(),
177			'fulfilled' => function (ResponseInterface $response, int $index) {
178			$this->handleResponse($response, $index);
179
180			$crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($index);
181
182			if ($crawlUrl->url->host !== $this->baseUrl->host) {
183			return;
184			}
185
186			$this->addAllLinksToCrawlQueue(
187			(string) $response->getBody(),
188			$crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($index)->url
189			);
190			},
191			'rejected' => function (RequestException $exception, int $index) {
192			$this->handleResponse($exception->getResponse(), $index);
193			},
194			]);
195
196			$promise = $pool->promise();
197			$promise->wait();
198
199			$this->crawlQueue->removeProcessedUrlsFromPending();
200			}
201			}
202
203			/**
204			* @param ResponseInterface\|null $response
205			* @param int $index
206			*/
207			protected function handleResponse($response, int $index)
208			{
209			$crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($index);
210
211			$this->crawlObserver->hasBeenCrawled($crawlUrl->url, $response, $crawlUrl->foundOnUrl);
212			}
213
214			protected function getCrawlRequests(): Generator
215			{
216			$i = 0;
217			while ($crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($i)) {
218			if (! $this->crawlProfile->shouldCrawl($crawlUrl->url)) {
219			$i++;
220			continue;
221			}
222
223			if ($this->crawlQueue->hasAlreadyBeenProcessed($crawlUrl)) {
224			$i++;
225			continue;
226			}
227
228			$this->crawlObserver->willCrawl($crawlUrl->url);
229
230			$this->crawlQueue->markAsProcessed($crawlUrl);
231
232			yield new Request('GET', (string) $crawlUrl->url);
233			$i++;
234			}
235			}
236
237			protected function addAllLinksToCrawlQueue(string $html, Url $foundOnUrl)
238			{
239			$allLinks = $this->extractAllLinks($html, $foundOnUrl);
240
241			collect($allLinks)
242			->filter(function (Url $url) {
243			return $url->hasCrawlableScheme();
244			})
245			->map(function (Url $url) use ($foundOnUrl) {
246			return $this->normalizeUrl($url);
247			})
248			->filter(function (Url $url) {
249			return $this->crawlProfile->shouldCrawl($url);
250			})
251			->reject(function ($url) {
252			return $this->crawlQueue->has($url);
253			})
254			->each(function (Url $url) use ($foundOnUrl) {
255			$node = $this->addToLinkTree($this->linkTree, (string)$url, $foundOnUrl);
256
257			if(($this->maximumDepth == 0) \|\| ($node->getDepth() <= $this->maximumDepth)) {
258			$this->crawlQueue->add(
259			CrawlUrl::create($url, $foundOnUrl)
260			);
261			}
262			});
263			}
264
265			protected function extractAllLinks(string $html, Url $foundOnUrl): Collection
266			{
267			if ($this->executeJavaScript) {
268			$html = $this->getBodyAfterExecutingJavaScript($foundOnUrl);
269			}
270
271			$domCrawler = new DomCrawler($html, $foundOnUrl);
272
273			return collect($domCrawler->filterXpath('//a')->links())
274			->map(function (Link $link) {
275			return Url::create($link->getUri());
276			});
277			}
278
279			protected function normalizeUrl(Url $url): Url
280			{
281			return $url->removeFragment();
282			}
283
284			/**
285			* @param $node \Tree\Node\Node
286			* @param $url string
287			* @param $parentUrl string
288			*/
289			protected function addToLinkTree(Node $node, string $url, string $parentUrl) {
290
291			$returnNode = null;
292			if($node->getValue() == $parentUrl) {
293			$newNode = new Node($url);
294			$node->addChild($newNode);
295
296			return $newNode;
297			}
298			foreach($node->getChildren() as $currentNode) {
299			$returnNode = $this->addToLinkTree($currentNode, $url, $parentUrl);
300			if($returnNode !== null) {
301			break;
302			}
303			}
304
305			return $returnNode;
306			}
307
308			protected function getBodyAfterExecutingJavaScript(Url $foundOnUrl): string
309			{
310			$browsershot = Browsershot::url((string) $foundOnUrl);
311
312			if ($this->pathToChromeBinary) {
			0 ignored issues – show Bug Best Practice introduced 2017-09-21 21:04 UTC by Report Bug Copy Issue Report The expression `$this->pathToChromeBinary` of type `string\|null` is loosely compared to `true`; this is ambiguous if the string can be empty. You might want to explicitly use `!== null` instead. In PHP, under loose comparison (like `==`, or `!=`, or `switch` conditions), values of different types might be equal. For `string` values, the empty string `''` is a special case, in particular the following results might be unexpected: '' == false // true '' == null // true 'ab' == false // false 'ab' == null // false // It is often better to use strict comparison '' === false // false '' === null // false Loading history...
313			$browsershot->setChromePath($this->pathToChromeBinary);
314			}
315
316			$html = $browsershot->bodyHtml();
317
318			return html_entity_decode($html);
319			}
320
321			}
322

spatie / crawler

Pull Request — master (#65)

Crawler::extractAllLinks() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like