|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Compolomus\RssReader; |
|
6
|
|
|
|
|
7
|
|
|
use DateTime; |
|
8
|
|
|
use DOMDocument; |
|
9
|
|
|
use DOMXPath; |
|
10
|
|
|
use SplFileObject; |
|
11
|
|
|
use Symfony\Component\Dotenv\Dotenv; |
|
12
|
|
|
|
|
13
|
|
|
use function PHPUnit\Framework\fileExists; |
|
14
|
|
|
|
|
15
|
|
|
class RssReader |
|
16
|
|
|
{ |
|
17
|
|
|
private SplFileObject $cache; |
|
18
|
|
|
|
|
19
|
|
|
private SplFileObject $cacheIds; |
|
20
|
|
|
|
|
21
|
|
|
private string $cacheFile; |
|
22
|
|
|
|
|
23
|
|
|
private string $cacheIdsFile; |
|
24
|
|
|
|
|
25
|
|
|
public function __construct(string $envFile = __DIR__ . '/.env') |
|
26
|
|
|
{ |
|
27
|
|
|
$dotenv = new Dotenv(); |
|
28
|
|
|
$dotenv->load($envFile); |
|
29
|
|
|
$dir = $_ENV['CACHEDIR']; |
|
30
|
|
|
|
|
31
|
|
|
$this->cacheFile = $dir . '/cacheChannels.txt'; |
|
32
|
|
|
$this->cacheIdsFile = $dir . '/cacheIds.txt'; |
|
33
|
|
|
|
|
34
|
|
|
|
|
35
|
|
|
if (!is_dir($dir)) { |
|
36
|
|
|
mkdir($dir); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
|
|
40
|
|
|
$this->cache = new SplFileObject($this->cacheFile, 'a+b'); |
|
41
|
|
|
$this->cacheIds = new SplFileObject($this->cacheIdsFile, 'a+b'); |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
public function addChannel(string $url): bool |
|
45
|
|
|
{ |
|
46
|
|
|
if (!in_array($url, file($this->cacheFile, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES), true)) { |
|
47
|
|
|
return (bool) $this->cache->fwrite($url . PHP_EOL); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
return false; |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
public function getAll(): array |
|
54
|
|
|
{ |
|
55
|
|
|
$dom = new DOMDocument(); |
|
56
|
|
|
$result = []; |
|
57
|
|
|
$ids = []; |
|
58
|
|
|
|
|
59
|
|
|
foreach ($this->getCacheChannels() as $chanel) { |
|
60
|
|
|
$dom->load($chanel); |
|
61
|
|
|
$items = $dom->getElementsByTagName('item'); |
|
62
|
|
|
$xpath = new DOMXpath($dom); |
|
63
|
|
|
foreach ($items as $item) { |
|
64
|
|
|
$link = $item->getElementsByTagName('link')->item(0)->nodeValue; |
|
65
|
|
|
preg_match('#\/(\d{3,})\/{0,1}#', $link, $matches); |
|
66
|
|
|
$timestamp = (new DateTime($item->getElementsByTagName('pubDate')->item(0)->nodeValue))->getTimestamp(); |
|
67
|
|
|
$id = $matches[1]; |
|
68
|
|
|
$cacheId = $id . '-' . $timestamp; |
|
69
|
|
|
if (!in_array($cacheId, $this->getCacheIds(), true)) { |
|
70
|
|
|
$result[] = [ |
|
71
|
|
|
'id' => $id, |
|
72
|
|
|
'title' => $item->getElementsByTagName('title')->item(0)->nodeValue, |
|
73
|
|
|
'desc' => trim(strip_tags($item->getElementsByTagName('description')->item(0)->nodeValue)), |
|
74
|
|
|
'link' => $link, |
|
75
|
|
|
'timestamp' => $timestamp, |
|
76
|
|
|
'img' => $xpath->query('//enclosure/@url')->item(0)->nodeValue, |
|
77
|
|
|
'cacheId' => $cacheId |
|
78
|
|
|
]; |
|
79
|
|
|
$ids[] = $cacheId; |
|
80
|
|
|
} |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
if (count($ids)) { |
|
85
|
|
|
$this->cacheIds->fwrite(implode(PHP_EOL, $ids) . PHP_EOL); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
return $result; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
protected function getCacheChannels(): array |
|
92
|
|
|
{ |
|
93
|
|
|
return fileExists($this->cacheFile) ? file( |
|
94
|
|
|
$this->cacheFile, |
|
95
|
|
|
FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES |
|
96
|
|
|
) : []; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
protected function getCacheIds(): array |
|
100
|
|
|
{ |
|
101
|
|
|
return fileExists($this->cacheIdsFile) ? file( |
|
102
|
|
|
$this->cacheIdsFile, |
|
103
|
|
|
FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES |
|
104
|
|
|
) : []; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
} |
|
108
|
|
|
|
Compolomus /
RssReader
