Crawler::shouldCrawl() - Code Metrics - Inspection of "Speed improvement. Not calculate the node depth, i..." - spatie/crawler - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#86)

by Peter

created 2017-12-10 12:36 UTC

Crawler::shouldCrawl() A

↳ Parent: Crawler

Complexity

Conditions	2
Paths	2

Size

Total Lines	8
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
c	0
b	0
f	0
dl	0
loc	8
rs	9.4285
cc	2
eloc	4
nc	2
nop	1

<?php

namespace Spatie\Crawler;

use Generator;
use Tree\Node\Node;
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;

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 $crawledUrlCount = 0;

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

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

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

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

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

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

    /**
     * @param array $clientOptions
     *
     * @return static
     */
    public static function create(array $clientOptions = [])
    {
        $clientOptions = (count($clientOptions))
            ? $clientOptions
            : self::$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 CrawlQueue();
    }

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

        return $this;
    }

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

        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);

        $this->addToCrawlQueue($crawlUrl);

        $this->depthTree = 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 (! $this->crawlProfile instanceof CrawlSubdomains) {
                        if ($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) {
                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 = $this->addtoDepthTree($this->depthTree, (string) $url, $foundOnUrl);

                if (! $this->shouldCrawl($node)) {
                    return;
                }

                if ($this->maximumCrawlCountReached()) {
                    return;
                }

                $crawlUrl = CrawlUrl::create($url, $foundOnUrl);

                $this->addToCrawlQueue($crawlUrl);
            });
    }

    protected function shouldCrawl(Node $node): bool
    {
        if (is_null($this->maximumDepth)) {
            return true;
        }

        return $node->getDepth() <= $this->maximumDepth;
    }

    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();
    }

    protected function addtoDepthTree(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 = $this->addtoDepthTree($currentNode, $url, $parentUrl);

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

    protected function addToCrawlQueue(CrawlUrl $crawlUrl)
    {
        $this->crawledUrlCount++;

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

        return $this;
    }

    protected function maximumCrawlCountReached(): bool
    {
        if (is_null($this->maximumCrawlCount)) {
            return false;
        }

        return $this->crawledUrlCount >= $this->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\Request;
10			use GuzzleHttp\RequestOptions;
11			use Illuminate\Support\Collection;
12			use Spatie\Browsershot\Browsershot;
13			use Symfony\Component\DomCrawler\Link;
14			use Psr\Http\Message\ResponseInterface;
15			use GuzzleHttp\Exception\RequestException;
16			use Symfony\Component\DomCrawler\Crawler as DomCrawler;
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 $crawledUrlCount = 0;
40
41			/** @var int\|null */
42			protected $maximumCrawlCount = null;
43
44			/** @var int\|null */
45			protected $maximumDepth = null;
46
47			/** @var \Tree\Node\Node */
48			protected $depthTree;
49
50			/** @var false */
51			protected $executeJavaScript = false;
52
53			/** @var string\|null */
54			protected $pathToChromeBinary = null;
55
56			protected static $defaultClientOptions = [
57			RequestOptions::COOKIES => true,
58			RequestOptions::CONNECT_TIMEOUT => 10,
59			RequestOptions::TIMEOUT => 10,
60			RequestOptions::ALLOW_REDIRECTS => false,
61			];
62
63			/**
64			* @param array $clientOptions
65			*
66			* @return static
67			*/
68			public static function create(array $clientOptions = [])
69			{
70			$clientOptions = (count($clientOptions))
71			? $clientOptions
72			: self::$defaultClientOptions;
73
74			$client = new Client($clientOptions);
75
76			return new static($client);
77			}
78
79			public function __construct(Client $client, int $concurrency = 10)
80			{
81			$this->client = $client;
82
83			$this->concurrency = $concurrency;
84
85			$this->crawlProfile = new CrawlAllUrls();
86
87			$this->crawlQueue = new CrawlQueue();
88			}
89
90			/**
91			* @param int $concurrency
92			*
93			* @return $this
94			*/
95			public function setConcurrency(int $concurrency)
96			{
97			$this->concurrency = $concurrency;
98
99			return $this;
100			}
101
102			/**
103			* @param int $maximumCrawlCount
104			*
105			* @return $this
106			*/
107			public function setMaximumCrawlCount(int $maximumCrawlCount)
108			{
109			$this->maximumCrawlCount = $maximumCrawlCount;
110
111			return $this;
112			}
113
114			/**
115			* @param int $maximumDepth
116			*
117			* @return $this
118			*/
119			public function setMaximumDepth(int $maximumDepth)
120			{
121			$this->maximumDepth = $maximumDepth;
122
123			return $this;
124			}
125
126			/**
127			* @return $this
128			*/
129			public function executeJavaScript($pathToChromeBinary = null)
130			{
131			$this->executeJavaScript = true;
			0 ignored issues – show Documentation Bug introduced 2017-09-21 20:55 UTC by Report Bug Copy Issue Report The property `$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...
132
133			$this->pathToChromeBinary = $pathToChromeBinary;
134
135			return $this;
136			}
137
138			/**
139			* @return $this
140			*/
141			public function doNotExecuteJavaScript()
142			{
143			$this->executeJavaScript = false;
144
145			return $this;
146			}
147
148			/**
149			* @param \Spatie\Crawler\CrawlObserver $crawlObserver
150			*
151			* @return $this
152			*/
153			public function setCrawlObserver(CrawlObserver $crawlObserver)
154			{
155			$this->crawlObserver = $crawlObserver;
156
157			return $this;
158			}
159
160			/**
161			* @param \Spatie\Crawler\CrawlProfile $crawlProfile
162			*
163			* @return $this
164			*/
165			public function setCrawlProfile(CrawlProfile $crawlProfile)
166			{
167			$this->crawlProfile = $crawlProfile;
168
169			return $this;
170			}
171
172			/**
173			* @param \Spatie\Crawler\Url\|string $baseUrl
174			*/
175			public function startCrawling($baseUrl)
176			{
177			if (! $baseUrl instanceof Url) {
178			$baseUrl = Url::create($baseUrl);
179			}
180
181			$this->baseUrl = $baseUrl;
182
183			$crawlUrl = CrawlUrl::create($baseUrl);
184
185			$this->addToCrawlQueue($crawlUrl);
186
187			$this->depthTree = new Node((string) $this->baseUrl);
188
189			$this->startCrawlingQueue();
190
191			$this->crawlObserver->finishedCrawling();
192			}
193
194			protected function startCrawlingQueue()
195			{
196			while ($this->crawlQueue->hasPendingUrls()) {
197			$pool = new Pool($this->client, $this->getCrawlRequests(), [
198			'concurrency' => $this->concurrency,
199			'options' => $this->client->getConfig(),
200			'fulfilled' => function (ResponseInterface $response, int $index) {
201			$this->handleResponse($response, $index);
202
203			$crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($index);
204
205			if (! $this->crawlProfile instanceof CrawlSubdomains) {
206			if ($crawlUrl->url->host !== $this->baseUrl->host) {
207			return;
208			}
209			}
210
211			$this->addAllLinksToCrawlQueue(
212			(string) $response->getBody(),
213			$crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($index)->url
214			);
215			},
216			'rejected' => function (RequestException $exception, int $index) {
217			$this->handleResponse($exception->getResponse(), $index);
218			},
219			]);
220
221			$promise = $pool->promise();
222			$promise->wait();
223
224			$this->crawlQueue->removeProcessedUrlsFromPending();
225			}
226			}
227
228			/**
229			* @param ResponseInterface\|null $response
230			* @param int $index
231			*/
232			protected function handleResponse($response, int $index)
233			{
234			$crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($index);
235
236			$this->crawlObserver->hasBeenCrawled($crawlUrl->url, $response, $crawlUrl->foundOnUrl);
237			}
238
239			protected function getCrawlRequests(): Generator
240			{
241			$i = 0;
242
243			while ($crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($i)) {
244			if (! $this->crawlProfile->shouldCrawl($crawlUrl->url)) {
245			$i++;
246			continue;
247			}
248
249			if ($this->crawlQueue->hasAlreadyBeenProcessed($crawlUrl)) {
250			$i++;
251			continue;
252			}
253
254			$this->crawlObserver->willCrawl($crawlUrl->url);
255
256			$this->crawlQueue->markAsProcessed($crawlUrl);
257
258			yield new Request('GET', (string) $crawlUrl->url);
259			$i++;
260			}
261			}
262
263			protected function addAllLinksToCrawlQueue(string $html, Url $foundOnUrl)
264			{
265			$allLinks = $this->extractAllLinks($html, $foundOnUrl);
266
267			collect($allLinks)
268			->filter(function (Url $url) {
269			return $url->hasCrawlableScheme();
270			})
271			->map(function (Url $url) {
272			return $this->normalizeUrl($url);
273			})
274			->filter(function (Url $url) {
275			return $this->crawlProfile->shouldCrawl($url);
276			})
277			->reject(function ($url) {
278			return $this->crawlQueue->has($url);
279			})
280			->each(function (Url $url) use ($foundOnUrl) {
281			$node = $this->addtoDepthTree($this->depthTree, (string) $url, $foundOnUrl);
282
283			if (! $this->shouldCrawl($node)) {
284			return;
285			}
286
287			if ($this->maximumCrawlCountReached()) {
288			return;
289			}
290
291			$crawlUrl = CrawlUrl::create($url, $foundOnUrl);
292
293			$this->addToCrawlQueue($crawlUrl);
294			});
295			}
296
297			protected function shouldCrawl(Node $node): bool
298			{
299			if (is_null($this->maximumDepth)) {
300			return true;
301			}
302
303			return $node->getDepth() <= $this->maximumDepth;
304			}
305
306			protected function extractAllLinks(string $html, Url $foundOnUrl): Collection
307			{
308			if ($this->executeJavaScript) {
309			$html = $this->getBodyAfterExecutingJavaScript($foundOnUrl);
310			}
311
312			$domCrawler = new DomCrawler($html, $foundOnUrl);
313
314			return collect($domCrawler->filterXpath('//a')->links())
315			->map(function (Link $link) {
316			return Url::create($link->getUri());
317			});
318			}
319
320			protected function normalizeUrl(Url $url): Url
321			{
322			return $url->removeFragment();
323			}
324
325			protected function addtoDepthTree(Node $node, string $url, string $parentUrl)
326			{
327			$returnNode = null;
328
329			if ($node->getValue() === $parentUrl) {
330			$newNode = new Node($url);
331
332			$node->addChild($newNode);
333
334			return $newNode;
335			}
336
337			foreach ($node->getChildren() as $currentNode) {
338			$returnNode = $this->addtoDepthTree($currentNode, $url, $parentUrl);
339
340			if (! is_null($returnNode)) {
341			break;
342			}
343			}
344
345			return $returnNode;
346			}
347
348			protected function getBodyAfterExecutingJavaScript(Url $foundOnUrl): string
349			{
350			$browsershot = Browsershot::url((string) $foundOnUrl);
351
352			if ($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...
353			$browsershot->setChromePath($this->pathToChromeBinary);
354			}
355
356			$html = $browsershot->bodyHtml();
357
358			return html_entity_decode($html);
359			}
360
361			protected function addToCrawlQueue(CrawlUrl $crawlUrl)
362			{
363			$this->crawledUrlCount++;
364
365			$this->crawlQueue->add($crawlUrl);
366
367			return $this;
368			}
369
370			protected function maximumCrawlCountReached(): bool
371			{
372			if (is_null($this->maximumCrawlCount)) {
373			return false;
374			}
375
376			return $this->crawledUrlCount >= $this->maximumCrawlCount;
377			}
378			}
379

spatie / crawler

Pull Request — master (#86)

Crawler::shouldCrawl() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like