|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of laravel.su package. |
|
4
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
5
|
|
|
* file that was distributed with this source code. |
|
6
|
|
|
*/ |
|
7
|
|
|
declare(strict_types=1); |
|
8
|
|
|
|
|
9
|
|
|
namespace App\Services\DataProviders; |
|
10
|
|
|
|
|
11
|
|
|
use Carbon\Carbon; |
|
12
|
|
|
use GuzzleHttp\Client; |
|
13
|
|
|
use Illuminate\Support\Collection; |
|
14
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Class LaravelNewsDataProvider. |
|
18
|
|
|
*/ |
|
19
|
|
|
class LaravelNewsDataProvider implements DataProviderInterface |
|
20
|
|
|
{ |
|
21
|
|
|
private const CONTENT_NAMESPACE = 'http://purl.org/rss/1.0/modules/content/'; |
|
22
|
|
|
private const FEED_URL = 'https://feed.laravel-news.com'; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var Client |
|
26
|
|
|
*/ |
|
27
|
|
|
private $client; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* LaravelNewsDataProvider constructor. |
|
31
|
|
|
* @param Client $client |
|
32
|
|
|
*/ |
|
33
|
|
|
public function __construct(Client $client) |
|
34
|
|
|
{ |
|
35
|
|
|
$this->client = $client; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @param \DateTime $latest |
|
40
|
|
|
* @return Collection|ExternalArticle[] |
|
41
|
|
|
* @throws \RuntimeException |
|
42
|
|
|
*/ |
|
43
|
|
|
public function getLatest(\DateTime $latest): Collection |
|
44
|
|
|
{ |
|
45
|
|
|
$result = new Collection(); |
|
46
|
|
|
|
|
47
|
|
|
foreach ($this->getArticles() as $article) { |
|
48
|
|
|
if (Carbon::instance($article->getCreatedAt())->addHour(1) < $latest) { |
|
49
|
|
|
break; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
$result->push($article); |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
return $result; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* @return \Generator|ExternalArticle[] |
|
60
|
|
|
* @throws \RuntimeException |
|
61
|
|
|
*/ |
|
62
|
|
|
private function getArticles(): \Generator |
|
63
|
|
|
{ |
|
64
|
|
|
$response = $this->client->get(self::FEED_URL); |
|
65
|
|
|
|
|
66
|
|
|
$body = $response->getBody(); |
|
67
|
|
|
|
|
68
|
|
|
yield from $this->parseBody((string) $body); |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* @param string $body |
|
73
|
|
|
* @return \Generator |
|
74
|
|
|
* @throws \RuntimeException |
|
75
|
|
|
*/ |
|
76
|
|
|
private function parseBody(string $body): \Generator |
|
77
|
|
|
{ |
|
78
|
|
|
$parser = new Crawler($body); |
|
79
|
|
|
|
|
80
|
|
|
/** @var \DOMElement $node */ |
|
81
|
|
|
foreach ($parser->filter('rss > channel > item') as $node) { |
|
82
|
|
|
$content = $node->getElementsByTagNameNS(self::CONTENT_NAMESPACE, '*') |
|
83
|
|
|
->item(0) |
|
84
|
|
|
->textContent; |
|
85
|
|
|
|
|
86
|
|
|
['images' => $images, 'body' => $content] = $this->parseContent( |
|
|
|
|
|
|
87
|
|
|
$content |
|
88
|
|
|
); |
|
89
|
|
|
|
|
90
|
|
|
['images' => $imagesPreview, 'body' => $preview] = $this->parseContent( |
|
|
|
|
|
|
91
|
|
|
(string) $this->getContentOf($node, 'description') |
|
92
|
|
|
); |
|
93
|
|
|
|
|
94
|
|
|
$title = (string) $this->getContentOf($node, 'title'); |
|
95
|
|
|
$link = (string) $this->getContentOf($node, 'link'); |
|
96
|
|
|
$published = (string) $this->getContentOf($node, 'pubDate'); |
|
97
|
|
|
|
|
98
|
|
|
$article = new ExternalArticle($title, trim($content), $link); |
|
99
|
|
|
|
|
100
|
|
|
$article->setPreview($preview); |
|
101
|
|
|
$article->setCreatedAt(Carbon::parse($published)); |
|
102
|
|
|
|
|
103
|
|
|
foreach ($images as $image) { |
|
104
|
|
|
$article->addImageUrl($image); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
yield $article; |
|
108
|
|
|
} |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* @param string $body |
|
113
|
|
|
* @return array |
|
114
|
|
|
*/ |
|
115
|
|
|
private function parseContent(string $body): array |
|
116
|
|
|
{ |
|
117
|
|
|
$images = []; |
|
118
|
|
|
|
|
119
|
|
|
$pattern = '/<img.+?src\s*=\s*"(.*?)".+?>/isu'; |
|
120
|
|
|
preg_match_all($pattern, $body, $matches); |
|
121
|
|
|
|
|
122
|
|
|
for ($i = 0, $len = count($matches[0]); $i < $len; $i++) { |
|
123
|
|
|
$images[] = $matches[1][$i] ?? null; |
|
124
|
|
|
$body = str_replace($matches[0][$i] ?? '', '', $body); |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
return ['images' => $images, 'body' => $body]; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
/** |
|
131
|
|
|
* @param \DOMElement $root |
|
132
|
|
|
* @param string $tagName |
|
133
|
|
|
* @return null|string |
|
134
|
|
|
*/ |
|
135
|
|
|
private function getContentOf(\DOMElement $root, string $tagName): ?string |
|
136
|
|
|
{ |
|
137
|
|
|
$node = $root->getElementsByTagName($tagName); |
|
138
|
|
|
|
|
139
|
|
|
if ($node->length >= 1) { |
|
140
|
|
|
return $node->item(0)->textContent; |
|
141
|
|
|
} |
|
142
|
|
|
} |
|
143
|
|
|
} |
|
144
|
|
|
|
LaravelRUS /
laravel.ru
Loading history...
This check marks access to variables or properties that have not been declared yet. While PHP has no explicit notion of declaring a variable, accessing it before a value is assigned to it is most likely a bug.