|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Dominikb\ComposerLicenseChecker; |
|
6
|
|
|
|
|
7
|
|
|
use DateTimeImmutable; |
|
8
|
|
|
use Dominikb\ComposerLicenseChecker\Contracts\LicenseLookup as LicenseLookupContract; |
|
9
|
|
|
use Dominikb\ComposerLicenseChecker\Exceptions\NoLookupPossibleException; |
|
10
|
|
|
use GuzzleHttp\ClientInterface; |
|
11
|
|
|
use GuzzleHttp\Exception\GuzzleException; |
|
12
|
|
|
use Symfony\Component\Cache\Adapter\FilesystemAdapter; |
|
13
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
14
|
|
|
use Symfony\Contracts\Cache\CacheInterface; |
|
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
|
4 |
|
public function __construct(ClientInterface $http, ?CacheInterface $cache = null) |
|
31
|
|
|
{ |
|
32
|
4 |
|
$this->http = $http; |
|
33
|
4 |
|
$this->cache = $cache ?? new FilesystemAdapter('FilesystemCache', 3600, __DIR__.'/../.cache'); |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
4 |
|
public function lookUp(string $licenseName): License |
|
37
|
|
|
{ |
|
38
|
4 |
|
return $this->cache->get($licenseName, function () use ($licenseName) { |
|
39
|
|
|
try { |
|
40
|
4 |
|
$detailsPageUrl = $this->queryForDetailPageUrl($licenseName); |
|
41
|
|
|
|
|
42
|
2 |
|
$license = $this->resolveLicenseInformation($licenseName, $detailsPageUrl); |
|
43
|
3 |
|
} catch (NoLookupPossibleException $exception) { |
|
44
|
3 |
|
$license = new NoLookupLicenses($licenseName); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
4 |
|
return $license; |
|
48
|
4 |
|
}); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* @param string $licenseShortName |
|
53
|
|
|
* @return string |
|
54
|
|
|
* |
|
55
|
|
|
* @throws NoLookupPossibleException |
|
56
|
|
|
*/ |
|
57
|
4 |
|
private function queryForDetailPageUrl(string $licenseShortName): string |
|
58
|
|
|
{ |
|
59
|
4 |
|
if (in_array($licenseShortName, self::$noLookup)) { |
|
60
|
1 |
|
throw new NoLookupPossibleException; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
3 |
|
$searchUrl = sprintf('%s/search?query=%s', static::API_HOST, $licenseShortName); |
|
64
|
|
|
|
|
65
|
|
|
try { |
|
66
|
3 |
|
$response = $this->http->request('get', $searchUrl); |
|
67
|
1 |
|
} catch (GuzzleException $exception) { |
|
68
|
1 |
|
throw new NoLookupPossibleException($exception->getMessage(), $exception->getCode(), $exception); |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
2 |
|
$crawler = $this->makeCrawler($response->getBody()->getContents()); |
|
72
|
|
|
|
|
73
|
2 |
|
$headings = $crawler->filter('div.search-result-items .cc-semibold')->extract(['_text']); |
|
74
|
2 |
|
$links = $crawler->filter('div.search-result-items .c-link-arrow')->extract(['href']); |
|
75
|
|
|
|
|
76
|
2 |
|
$zipped = array_map(null, $headings, $links); |
|
|
|
|
|
|
77
|
|
|
|
|
78
|
2 |
|
$relativeUrl = $this->findBestMatch($zipped, $licenseShortName); |
|
79
|
|
|
|
|
80
|
2 |
|
return static::API_HOST.$relativeUrl; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
2 |
|
private function makeCrawler(string $html): Crawler |
|
84
|
|
|
{ |
|
85
|
2 |
|
return new Crawler($html); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* @param string $licenseShortName |
|
90
|
|
|
* @param string $detailsPageUrl |
|
91
|
|
|
* @return License |
|
92
|
|
|
* |
|
93
|
|
|
* @throws NoLookupPossibleException |
|
94
|
|
|
*/ |
|
95
|
2 |
|
private function resolveLicenseInformation(string $licenseShortName, string $detailsPageUrl): License |
|
96
|
|
|
{ |
|
97
|
|
|
try { |
|
98
|
2 |
|
$response = $this->http->request('get', $detailsPageUrl); |
|
99
|
1 |
|
$pageContent = $response->getBody()->getContents(); |
|
100
|
|
|
|
|
101
|
1 |
|
$crawler = $this->makeCrawler($pageContent); |
|
102
|
|
|
|
|
103
|
1 |
|
$license = (new License($licenseShortName)) |
|
104
|
1 |
|
->setCan($this->extractCans($crawler)) |
|
105
|
1 |
|
->setCannot($this->extractCannots($crawler)) |
|
106
|
1 |
|
->setMust($this->extractMusts($crawler)) |
|
107
|
1 |
|
->setSource($detailsPageUrl) |
|
108
|
1 |
|
->setCreatedAt(new DateTimeImmutable); |
|
109
|
|
|
|
|
110
|
1 |
|
return $license; |
|
111
|
1 |
|
} catch (GuzzleException $exception) { |
|
112
|
1 |
|
throw new NoLookupPossibleException($exception->getMessage(), $exception->getCode(), $exception); |
|
113
|
|
|
} |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
1 |
|
private function extractCans(Crawler $crawler): array |
|
117
|
|
|
{ |
|
118
|
1 |
|
return $this->extractListByIndex($crawler, 1); |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
1 |
|
private function extractCannots(Crawler $crawler): array |
|
122
|
|
|
{ |
|
123
|
1 |
|
return $this->extractListByIndex($crawler, 3); |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
1 |
|
private function extractMusts(Crawler $crawler): array |
|
127
|
|
|
{ |
|
128
|
1 |
|
return $this->extractListByIndex($crawler, 5); |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
// Index is the offset as children of div.features |
|
132
|
1 |
|
private function extractListByIndex(Crawler $crawler, $index): array |
|
133
|
|
|
{ |
|
134
|
1 |
|
$headings = $crawler->filter('div.c-feature:nth-child('.$index.') .c-text-md.cc-semibold') |
|
135
|
1 |
|
->each(function (Crawler $crawler) { |
|
136
|
|
|
return $crawler->getNode(0)->textContent; |
|
137
|
1 |
|
}); |
|
138
|
|
|
|
|
139
|
1 |
|
$bodies = $crawler->filter('div.c-feature:nth-child('.$index.') .c-text-sm') |
|
140
|
1 |
|
->each(function (Crawler $crawler) { |
|
141
|
|
|
return $crawler->getNode(0)->textContent; |
|
142
|
1 |
|
}); |
|
143
|
|
|
|
|
144
|
1 |
|
return array_combine($headings, $bodies); |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* Find the best matching link by comparing the similarity of the link and text. |
|
149
|
|
|
* |
|
150
|
|
|
* @param array $zipped |
|
151
|
|
|
* @param string $licenseShortName |
|
152
|
|
|
* @return string |
|
153
|
|
|
*/ |
|
154
|
2 |
|
private function findBestMatch(array $zipped, string $licenseShortName): string |
|
155
|
|
|
{ |
|
156
|
2 |
|
$bestMatch = 0; |
|
157
|
2 |
|
$matchingLink = ''; |
|
158
|
|
|
|
|
159
|
2 |
|
foreach ($zipped as [$title, $link]) { |
|
160
|
2 |
|
$titleMatch = similar_text($title, $licenseShortName); |
|
161
|
2 |
|
$linkMatch = similar_text($link, $licenseShortName); |
|
162
|
|
|
|
|
163
|
2 |
|
$totalMatch = $titleMatch + $linkMatch; |
|
164
|
|
|
|
|
165
|
2 |
|
if ($totalMatch > $bestMatch) { |
|
166
|
2 |
|
$bestMatch = $totalMatch; |
|
167
|
2 |
|
$matchingLink = $link; |
|
168
|
|
|
} |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
2 |
|
return $matchingLink; |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
|
|
public function setCache(CacheInterface $cache): void |
|
175
|
|
|
{ |
|
176
|
|
|
$this->cache = $cache; |
|
177
|
|
|
} |
|
178
|
|
|
} |
|
179
|
|
|
|
dominikb /
composer-license-checker
Loading history...