LicenseLookup::makeCrawler() - Code Metrics - Inspection of "wip" - dominikb/composer-license-checker - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( d33d35...37f9e4 )

by Dominik

created 2019-04-06 17:32 UTC

LicenseLookup::makeCrawler() A

↳ Parent: LicenseLookup

Complexity

Conditions	1
Paths	1

Size

Total Lines	3
Code Lines	1

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	1
dl	0
loc	3
rs	10
c	0
b	0
f	0
cc	1
nc	1
nop	1

<?php

declare(strict_types=1);

namespace Dominikb\ComposerLicenseChecker;

use DateTimeImmutable;
use Dominikb\ComposerLicenseChecker\Exceptions\NoLookupPossibleException;
use GuzzleHttp\Exception\GuzzleException;
use GuzzleHttp\ClientInterface;
use Psr\SimpleCache\CacheInterface;
use Symfony\Component\DomCrawler\Crawler;
use Symfony\Component\Cache\Simple\FilesystemCache;
use Dominikb\ComposerLicenseChecker\Contracts\LicenseLookup as LicenseLookupContract;

class LicenseLookup implements LicenseLookupContract
{
    const API_HOST = 'https://tldrlegal.com';

    /** @var ClientInterface */
    protected $http;
    /** @var CacheInterface */
    protected $cache;
    /** @var string[] */
    private static $noLookup = [
        'none',
        'proprietary',
    ];

    public function __construct(ClientInterface $http, CacheInterface $cache = null)
    {
        $this->http = $http;
        $this->cache = $cache ?? new FilesystemCache('LicenseLookup', 3600, __DIR__.'/../.cache');
    }

    public function lookUp(string $licenseName): License
    {
        if ($cached = $this->cache->get($licenseName)) {
            return $cached;
        }

        try {
            $detailsPageUrl = $this->queryForDetailPageUrl($licenseName);

            $license = $this->resolveLicenseInformation($licenseName, $detailsPageUrl);
        }catch (NoLookupPossibleException $exception) {
            $license = new NoLookupLicenses($licenseName);
        }

        $this->cache->set($licenseName, $license);

        return $license;
    }

    /**
     * @param string $licenseShortName
     *
     * @return string
     * @throws NoLookupPossibleException
     */
    private function queryForDetailPageUrl(string $licenseShortName): string
    {
        if (in_array($licenseShortName, static::$noLookup)) {

            throw new NoLookupPossibleException;
        }

        $searchUrl = sprintf('%s/search?q=%s', static::API_HOST, $licenseShortName);

        try {
            $response = $this->http->request('get', $searchUrl);
        } catch (GuzzleException $exception) {
            throw new NoLookupPossibleException($exception->getMessage(), $exception->getCode(), $exception);
        }

        $crawler = $this->makeCrawler($response->getBody()->getContents());

        $headings = $crawler->filter('div#licenses > .search-result > a > h3')->extract(['_text']);
        $links = $crawler->filter('div#licenses > .search-result > a')->extract(['href']);

        $zipped = array_map(null, $headings, $links);

        $relativeUrl = $this->findBestMatch($zipped, $licenseShortName);

        return static::API_HOST.$relativeUrl;
    }

    private function makeCrawler(string $html): Crawler
    {
        return new Crawler($html);
    }

    /**
     * @param string $licenseShortName
     * @param string $detailsPageUrl
     *
     * @return License
     * @throws NoLookupPossibleException
     */
    private function resolveLicenseInformation(string $licenseShortName, string $detailsPageUrl): License
    {
        try {
            $response = $this->http->request('get', $detailsPageUrl);
            $pageContent = $response->getBody()->getContents();

            $crawler = $this->makeCrawler($pageContent);

            $license = (new License($licenseShortName))
                ->setCan($this->extractCans($crawler))
                ->setCannot($this->extractCannots($crawler))
                ->setMust($this->extractMusts($crawler))
                ->setSource($detailsPageUrl)
                ->setCreatedAt(new DateTimeImmutable);

            return $license;
        } catch (GuzzleException $exception) {
            throw new NoLookupPossibleException($exception->getMessage(), $exception->getCode(), $exception);
        }
    }

    private function extractCans(Crawler $crawler): array
    {
        return $this->extractListByColor($crawler, 'green');
    }

    private function extractCannots(Crawler $crawler): array
    {
        return $this->extractListByColor($crawler, 'red');
    }

    private function extractMusts(Crawler $crawler): array
    {
        return $this->extractListByColor($crawler, 'blue');
    }

    private function extractListByColor(Crawler $crawler, $color): array
    {
        $headings = $crawler->filter(".bucket-list.$color li div.attr-head")
                            ->each(function (Crawler $crawler) {
                                return $crawler->getNode(0)->textContent;
                            });

        $bodies = $crawler->filter(".bucket-list.$color li div.attr-body")
                          ->each(function (Crawler $crawler) {
                              return $crawler->getNode(0)->textContent;
                          });

        return array_combine($headings, $bodies);
    }

    /**
     * Find the best matching link by comparing the similarity of the link and text.
     *
     * @param array  $zipped
     * @param string $licenseShortName
     *
     * @return string
     */
    private function findBestMatch(array $zipped, string $licenseShortName): string
    {
        $bestMatch = 0;
        $matchingLink = '';

        foreach($zipped as [$title, $link]) {
            $titleMatch = similar_text($title, $licenseShortName);
            $linkMatch = similar_text($link, $licenseShortName);

            $totalMatch = $titleMatch + $linkMatch;

            if ($totalMatch > $bestMatch) {
                $bestMatch = $totalMatch;
                $matchingLink = $link;
            }
        }

        return $matchingLink;
    }

    public function setCache(CacheInterface $cache): void
    {
        $this->cache = $cache;
    }
}


1			<?php
2
3			declare(strict_types=1);
4
5			namespace Dominikb\ComposerLicenseChecker;
6
7			use DateTimeImmutable;
8			use Dominikb\ComposerLicenseChecker\Exceptions\NoLookupPossibleException;
9			use GuzzleHttp\Exception\GuzzleException;
10			use GuzzleHttp\ClientInterface;
11			use Psr\SimpleCache\CacheInterface;
12			use Symfony\Component\DomCrawler\Crawler;
13			use Symfony\Component\Cache\Simple\FilesystemCache;
14			use Dominikb\ComposerLicenseChecker\Contracts\LicenseLookup as LicenseLookupContract;
15
16			class LicenseLookup implements LicenseLookupContract
17			{
18			const API_HOST = 'https://tldrlegal.com';
19
20			/** @var ClientInterface */
21			protected $http;
22			/** @var CacheInterface */
23			protected $cache;
24			/** @var string[] */
25			private static $noLookup = [
26			'none',
27			'proprietary',
28			];
29
30			public function __construct(ClientInterface $http, CacheInterface $cache = null)
31			{
32			$this->http = $http;
33			$this->cache = $cache ?? new FilesystemCache('LicenseLookup', 3600, __DIR__.'/../.cache');
34			}
35
36			public function lookUp(string $licenseName): License
37			{
38			if ($cached = $this->cache->get($licenseName)) {
39			return $cached;
40			}
41
42			try {
43			$detailsPageUrl = $this->queryForDetailPageUrl($licenseName);
44
45			$license = $this->resolveLicenseInformation($licenseName, $detailsPageUrl);
46			}catch (NoLookupPossibleException $exception) {
47			$license = new NoLookupLicenses($licenseName);
48			}
49
50			$this->cache->set($licenseName, $license);
51
52			return $license;
53			}
54
55			/**
56			* @param string $licenseShortName
57			*
58			* @return string
59			* @throws NoLookupPossibleException
60			*/
61			private function queryForDetailPageUrl(string $licenseShortName): string
62			{
63			if (in_array($licenseShortName, static::$noLookup)) {
			0 ignored issues – show Bug introduced 2019-04-06 17:33 UTC by Report Bug Copy Issue Report Since `$noLookup` is declared private, accessing it with `static` will lead to errors in possible sub-classes; you can either use `self`, or increase the visibility of `$noLookup` to at least protected. Loading history...
64			throw new NoLookupPossibleException;
65			}
66
67			$searchUrl = sprintf('%s/search?q=%s', static::API_HOST, $licenseShortName);
68
69			try {
70			$response = $this->http->request('get', $searchUrl);
71			} catch (GuzzleException $exception) {
72			throw new NoLookupPossibleException($exception->getMessage(), $exception->getCode(), $exception);
73			}
74
75			$crawler = $this->makeCrawler($response->getBody()->getContents());
76
77			$headings = $crawler->filter('div#licenses > .search-result > a > h3')->extract(['_text']);
78			$links = $crawler->filter('div#licenses > .search-result > a')->extract(['href']);
79
80			$zipped = array_map(null, $headings, $links);
81
82			$relativeUrl = $this->findBestMatch($zipped, $licenseShortName);
83
84			return static::API_HOST.$relativeUrl;
85			}
86
87			private function makeCrawler(string $html): Crawler
88			{
89			return new Crawler($html);
90			}
91
92			/**
93			* @param string $licenseShortName
94			* @param string $detailsPageUrl
95			*
96			* @return License
97			* @throws NoLookupPossibleException
98			*/
99			private function resolveLicenseInformation(string $licenseShortName, string $detailsPageUrl): License
100			{
101			try {
102			$response = $this->http->request('get', $detailsPageUrl);
103			$pageContent = $response->getBody()->getContents();
104
105			$crawler = $this->makeCrawler($pageContent);
106
107			$license = (new License($licenseShortName))
108			->setCan($this->extractCans($crawler))
109			->setCannot($this->extractCannots($crawler))
110			->setMust($this->extractMusts($crawler))
111			->setSource($detailsPageUrl)
112			->setCreatedAt(new DateTimeImmutable);
113
114			return $license;
115			} catch (GuzzleException $exception) {
116			throw new NoLookupPossibleException($exception->getMessage(), $exception->getCode(), $exception);
117			}
118			}
119
120			private function extractCans(Crawler $crawler): array
121			{
122			return $this->extractListByColor($crawler, 'green');
123			}
124
125			private function extractCannots(Crawler $crawler): array
126			{
127			return $this->extractListByColor($crawler, 'red');
128			}
129
130			private function extractMusts(Crawler $crawler): array
131			{
132			return $this->extractListByColor($crawler, 'blue');
133			}
134
135			private function extractListByColor(Crawler $crawler, $color): array
136			{
137			$headings = $crawler->filter(".bucket-list.$color li div.attr-head")
138			->each(function (Crawler $crawler) {
139			return $crawler->getNode(0)->textContent;
140			});
141
142			$bodies = $crawler->filter(".bucket-list.$color li div.attr-body")
143			->each(function (Crawler $crawler) {
144			return $crawler->getNode(0)->textContent;
145			});
146
147			return array_combine($headings, $bodies);
148			}
149
150			/**
151			* Find the best matching link by comparing the similarity of the link and text.
152			*
153			* @param array $zipped
154			* @param string $licenseShortName
155			*
156			* @return string
157			*/
158			private function findBestMatch(array $zipped, string $licenseShortName): string
159			{
160			$bestMatch = 0;
161			$matchingLink = '';
162
163			foreach($zipped as [$title, $link]) {
164			$titleMatch = similar_text($title, $licenseShortName);
165			$linkMatch = similar_text($link, $licenseShortName);
166
167			$totalMatch = $titleMatch + $linkMatch;
168
169			if ($totalMatch > $bestMatch) {
170			$bestMatch = $totalMatch;
171			$matchingLink = $link;
172			}
173			}
174
175			return $matchingLink;
176			}
177
178			public function setCache(CacheInterface $cache): void
179			{
180			$this->cache = $cache;
181			}
182			}
183

dominikb / composer-license-checker

Push — master ( d33d35...37f9e4 )

LicenseLookup::makeCrawler() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like