Diglett::getUrl() - Code Metrics - Inspection of "Add fetching urls without selectors" - jerodev/diglett - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 6a5c93...71d76d )

by Jeroen

created 2019-06-19 15:17 UTC

Diglett::getUrl() A

↳ Parent: Diglett

Complexity

Conditions	6
Paths	9

Size

Total Lines	27
Code Lines	18

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	6
eloc	18
nc	9
nop	0
dl	0
loc	27
rs	9.0444
c	0
b	0
f	0

<?php

namespace Jerodev\Diglett;

use Closure;
use Jerodev\Diglett\CssFilters\ICssFilter;
use Symfony\Component\DomCrawler\Crawler;

class Diglett
{
    /**
     *  The Symfony DomCrawler to work with.
     *
     *  @var Crawler|null
     */
    private $crawler;

    /**
     *  The css selector parser.
     *
     *  @var CssFilterParser
     */
    private $cssFilterParser;

    /**
     *  Create a diglett instance from a Symfony Crawler.
     *
     *  @param Crawler|null $crawler
     *  @param ICssFilter[] $cssFilters An array of extra ICssFilterl classes to filter on
     */
    public function __construct(?Crawler $crawler = null, array $cssFilters = [])
    {
        $this->crawler = $crawler;
        $this->cssFilterParser = new CssFilterParser($cssFilters);
    }

    /**
     *  Get the underlying crawler object.
     *
     *  @return Crawler|null
     */
    public function getCrawler(): ?Crawler
    {
        return $this->crawler;
    }

    /**
     *  Perform a closure function on matched nodes for a selector and return as array.
     *
     *  @param string $selector
     *  @param Closure $closure A function to perform on the list of nodes
     *
     *  @return array An array of results returned by the closure
     */
    public function each(string $selector, Closure $closure): array
    {
        return $this->filter($selector)->getCrawler()->each(function ($crawler, $i) use ($closure) {
            return $closure(new self($crawler), $i);
        });
    }

    /**
     *  Use special css selectors to filter on the current node collection.
     *
     *  @param string $selector
     *
     *  @return Diglett
     */
    public function filter(string $selector): self
    {
        $parsedSelector = $this->cssFilterParser->parse($selector);

        $crawler = $this->getCrawler();
        foreach ($parsedSelector as $part) {
            if (empty($crawler) || $crawler->count() === 0) {
                break;
            }

            if (!empty($part->getSelector())) {
                $crawler = $crawler->filter($part->getSelector());
            }

            foreach ($part->getFunctions() as $function) {
                $crawler = $function->filterNodes($crawler);
                if ($crawler === null) {
                    break;
                }
            }
        }

        return new self($crawler);
    }

    /**
     *  Use special css selectors to fetch several values.
     *
     *  @param string[] $selectors
     *
     *  @return array
     */
    public function getTexts(array $selectors): array
    {
        $results = [];
        foreach ($selectors as $key => $value) {
            $results[$key] = $this->getText($value);
        }

        return $results;
    }

    /**
     *  Get the value for a single special css selector.
     *
     *  @param string $selector
     *
     *  @return string|null
     */
    public function getText(?string $selector = null): ?string
    {
        $attribute = null;
        $diglett = $this;

        if ($selector !== null) {
            if (($attr = strstr($selector, '{')) && $attr[-1] === '}') {
                $selector = substr($selector, 0, strlen($attr) * -1);
                $attribute = substr($attr, 1, -1);
            }

            $diglett = $this->filter($selector);
        }

        if ($diglett->nodeCount() === 0) {
            return null;
        }

        $crawler = $diglett->getCrawler();

        return $attribute === null ? $crawler->text() : $crawler->attr($attribute);
    }

    /**
     *  Fetch urls from the selected nodes (a[href], img[src]).
     */
    public function getUrls(?string $selector = null): array
    {
        if ($selector !== null) {
            $diglett = $this->filter($selector);
        } else {
            $diglett = new self($this->getCrawler());
        }
        
        if ($diglett->nodeCount() === 0) {
            return [];
        }

        $urls = $diglett->each('a, img', function ($d) {
            return $d->getUrl();
        });

        return array_filter($urls);
    }

    /**
     *  Fetch the url from the current main node if available.
     */
    public function getUrl(): ?string
    {
        $crawler = $this->getCrawler();
        $absolute = implode('/', array_slice(explode('/', $crawler->getUri()), 0, 3)) . '/';
        $relative = substr(strstr($crawler->getUri(), '?', true) ?: $crawler->getUri(), 0, strrpos($crawler->getUri(), '/') + 1);

        $url = null;
        $node = $crawler->first();
        switch (strtolower($node->nodeName())) {
            case 'a':
                $url = $node->attr('href');
                break;

            case 'img':
                $url = $node->attr('src');
                break;
        }

        if (!in_array(substr($url, 0, 7), ['http://', 'https:/'])) {
            if ($url[0] === '/') {
                $url = $absolute . ltrim($url, '/');
            } else {
                $url = $relative . ltrim($url, '/');
            }
        }

        return $url;
    }

    /**
     *  Find the node count on the current crawler instance.
     */
    public function nodeCount(): int
    {
        if ($this->crawler === null) {
            return 0;
        }

        return $this->crawler->count();
    }
}


1			<?php
2
3			namespace Jerodev\Diglett;
4
5			use Closure;
6			use Jerodev\Diglett\CssFilters\ICssFilter;
7			use Symfony\Component\DomCrawler\Crawler;
8
9			class Diglett
10			{
11			/**
12			* The Symfony DomCrawler to work with.
13			*
14			* @var Crawler\|null
15			*/
16			private $crawler;
17
18			/**
19			* The css selector parser.
20			*
21			* @var CssFilterParser
22			*/
23			private $cssFilterParser;
24
25			/**
26			* Create a diglett instance from a Symfony Crawler.
27			*
28			* @param Crawler\|null $crawler
29			* @param ICssFilter[] $cssFilters An array of extra ICssFilterl classes to filter on
30			*/
31			public function __construct(?Crawler $crawler = null, array $cssFilters = [])
32			{
33			$this->crawler = $crawler;
34			$this->cssFilterParser = new CssFilterParser($cssFilters);
35			}
36
37			/**
38			* Get the underlying crawler object.
39			*
40			* @return Crawler\|null
41			*/
42			public function getCrawler(): ?Crawler
43			{
44			return $this->crawler;
45			}
46
47			/**
48			* Perform a closure function on matched nodes for a selector and return as array.
49			*
50			* @param string $selector
51			* @param Closure $closure A function to perform on the list of nodes
52			*
53			* @return array An array of results returned by the closure
54			*/
55			public function each(string $selector, Closure $closure): array
56			{
57			return $this->filter($selector)->getCrawler()->each(function ($crawler, $i) use ($closure) {
58			return $closure(new self($crawler), $i);
59			});
60			}
61
62			/**
63			* Use special css selectors to filter on the current node collection.
64			*
65			* @param string $selector
66			*
67			* @return Diglett
68			*/
69			public function filter(string $selector): self
70			{
71			$parsedSelector = $this->cssFilterParser->parse($selector);
72
73			$crawler = $this->getCrawler();
74			foreach ($parsedSelector as $part) {
75			if (empty($crawler) \|\| $crawler->count() === 0) {
76			break;
77			}
78
79			if (!empty($part->getSelector())) {
80			$crawler = $crawler->filter($part->getSelector());
81			}
82
83			foreach ($part->getFunctions() as $function) {
84			$crawler = $function->filterNodes($crawler);
85			if ($crawler === null) {
86			break;
87			}
88			}
89			}
90
91			return new self($crawler);
92			}
93
94			/**
95			* Use special css selectors to fetch several values.
96			*
97			* @param string[] $selectors
98			*
99			* @return array
100			*/
101			public function getTexts(array $selectors): array
102			{
103			$results = [];
104			foreach ($selectors as $key => $value) {
105			$results[$key] = $this->getText($value);
106			}
107
108			return $results;
109			}
110
111			/**
112			* Get the value for a single special css selector.
113			*
114			* @param string $selector
115			*
116			* @return string\|null
117			*/
118			public function getText(?string $selector = null): ?string
119			{
120			$attribute = null;
121			$diglett = $this;
122
123			if ($selector !== null) {
124			if (($attr = strstr($selector, '{')) && $attr[-1] === '}') {
125			$selector = substr($selector, 0, strlen($attr) * -1);
126			$attribute = substr($attr, 1, -1);
127			}
128
129			$diglett = $this->filter($selector);
130			}
131
132			if ($diglett->nodeCount() === 0) {
133			return null;
134			}
135
136			$crawler = $diglett->getCrawler();
137
138			return $attribute === null ? $crawler->text() : $crawler->attr($attribute);
139			}
140
141			/**
142			* Fetch urls from the selected nodes (a[href], img[src]).
143			*/
144			public function getUrls(?string $selector = null): array
145			{
146			if ($selector !== null) {
147			$diglett = $this->filter($selector);
148			} else {
149			$diglett = new self($this->getCrawler());
150			}
151
152			if ($diglett->nodeCount() === 0) {
153			return [];
154			}
155
156			$urls = $diglett->each('a, img', function ($d) {
157			return $d->getUrl();
158			});
159
160			return array_filter($urls);
161			}
162
163			/**
164			* Fetch the url from the current main node if available.
165			*/
166			public function getUrl(): ?string
167			{
168			$crawler = $this->getCrawler();
169			$absolute = implode('/', array_slice(explode('/', $crawler->getUri()), 0, 3)) . '/';
170			$relative = substr(strstr($crawler->getUri(), '?', true) ?: $crawler->getUri(), 0, strrpos($crawler->getUri(), '/') + 1);
171
172			$url = null;
173			$node = $crawler->first();
174			switch (strtolower($node->nodeName())) {
175			case 'a':
176			$url = $node->attr('href');
177			break;
178
179			case 'img':
180			$url = $node->attr('src');
181			break;
182			}
183
184			if (!in_array(substr($url, 0, 7), ['http://', 'https:/'])) {
185			if ($url[0] === '/') {
186			$url = $absolute . ltrim($url, '/');
187			} else {
188			$url = $relative . ltrim($url, '/');
189			}
190			}
191
192			return $url;
193			}
194
195			/**
196			* Find the node count on the current crawler instance.
197			*/
198			public function nodeCount(): int
199			{
200			if ($this->crawler === null) {
201			return 0;
202			}
203
204			return $this->crawler->count();
205			}
206			}
207

jerodev / diglett

Push — master ( 6a5c93...71d76d )

Diglett::getUrl() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like