|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types = 1); |
|
3
|
|
|
|
|
4
|
|
|
namespace PersonalGalaxy\RSS\Parser; |
|
5
|
|
|
|
|
6
|
|
|
use PersonalGalaxy\RSS\{ |
|
7
|
|
|
Entity\Article, |
|
8
|
|
|
Entity\Article\Author, |
|
9
|
|
|
Entity\Article\Description, |
|
10
|
|
|
Entity\Article\Title, |
|
11
|
|
|
UrlFactory, |
|
12
|
|
|
}; |
|
13
|
|
|
use Innmind\Url\Url; |
|
14
|
|
|
use Innmind\Crawler\{ |
|
15
|
|
|
Parser, |
|
16
|
|
|
HttpResource\Attribute\Attribute, |
|
17
|
|
|
}; |
|
18
|
|
|
use Innmind\Http\Message\{ |
|
19
|
|
|
Request, |
|
20
|
|
|
Response, |
|
21
|
|
|
}; |
|
22
|
|
|
use Innmind\Xml\{ |
|
23
|
|
|
ReaderInterface, |
|
24
|
|
|
NodeInterface, |
|
25
|
|
|
ElementInterface, |
|
26
|
|
|
}; |
|
27
|
|
|
use Innmind\TimeContinuum\TimeContinuumInterface; |
|
28
|
|
|
use Innmind\Immutable\{ |
|
29
|
|
|
MapInterface, |
|
30
|
|
|
SetInterface, |
|
31
|
|
|
Set, |
|
32
|
|
|
StreamInterface, |
|
33
|
|
|
Stream, |
|
34
|
|
|
}; |
|
35
|
|
|
|
|
36
|
|
|
final class RSS implements Parser |
|
37
|
|
|
{ |
|
38
|
|
|
private $reader; |
|
39
|
|
|
private $clock; |
|
40
|
|
|
private $make; |
|
41
|
|
|
|
|
42
|
4 |
|
public function __construct( |
|
43
|
|
|
ReaderInterface $reader, |
|
44
|
|
|
TimeContinuumInterface $clock, |
|
45
|
|
|
UrlFactory $make |
|
46
|
|
|
) { |
|
47
|
4 |
|
$this->reader = $reader; |
|
48
|
4 |
|
$this->clock = $clock; |
|
49
|
4 |
|
$this->make = $make; |
|
50
|
4 |
|
} |
|
51
|
|
|
|
|
52
|
3 |
|
public function parse( |
|
53
|
|
|
Request $request, |
|
54
|
|
|
Response $response, |
|
55
|
|
|
MapInterface $attributes |
|
56
|
|
|
): MapInterface { |
|
57
|
|
|
if ( |
|
58
|
3 |
|
!$response->headers()->has('content-type') || |
|
59
|
3 |
|
(string) $response->headers()->get('content-type')->values()->current() !== 'application/rss' |
|
60
|
|
|
) { |
|
61
|
2 |
|
return $attributes; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
1 |
|
$xml = $this->reader->read($response->body()); |
|
65
|
|
|
$articles = $xml |
|
66
|
1 |
|
->children() |
|
67
|
1 |
|
->values() |
|
68
|
1 |
|
->first() // rss node |
|
69
|
1 |
|
->children() |
|
70
|
1 |
|
->values() |
|
71
|
1 |
|
->filter(static function(NodeInterface $node): bool { |
|
72
|
1 |
|
return $node instanceof ElementInterface && $node->name() === 'channel'; |
|
73
|
1 |
|
}) |
|
74
|
1 |
|
->reduce( |
|
75
|
1 |
|
Stream::of(NodeInterface::class), |
|
76
|
1 |
|
static function(StreamInterface $channels, ElementInterface $channel): StreamInterface { |
|
77
|
1 |
|
return $channels->append($channel->children()->values()); |
|
78
|
1 |
|
} |
|
79
|
|
|
) |
|
80
|
1 |
|
->filter(static function(NodeInterface $node): bool { |
|
81
|
1 |
|
return $node instanceof ElementInterface && $node->name() === 'item'; |
|
82
|
1 |
|
}) |
|
83
|
1 |
|
->reduce( |
|
84
|
1 |
|
Set::of(Article::class), |
|
85
|
1 |
|
function(SetInterface $articles, ElementInterface $item): SetInterface { |
|
86
|
|
|
$elements = $item |
|
87
|
1 |
|
->children() |
|
88
|
1 |
|
->values() |
|
89
|
1 |
|
->filter(static function(NodeInterface $node): bool { |
|
90
|
1 |
|
return $node instanceof ElementInterface; |
|
91
|
1 |
|
}); |
|
92
|
|
|
$link = $elements |
|
93
|
1 |
|
->filter(static function(ElementInterface $element): bool { |
|
94
|
1 |
|
return $element->name() === 'link'; |
|
95
|
1 |
|
}) |
|
96
|
1 |
|
->first() |
|
97
|
1 |
|
->content(); |
|
98
|
|
|
$author = $elements |
|
99
|
1 |
|
->filter(static function(ElementInterface $element): bool { |
|
100
|
1 |
|
return $element->name() === 'author' || $element->name() === 'dc:creator'; |
|
101
|
1 |
|
}) |
|
102
|
1 |
|
->first() |
|
103
|
1 |
|
->content(); |
|
104
|
|
|
$description = $elements |
|
105
|
1 |
|
->filter(static function(ElementInterface $element): bool { |
|
106
|
1 |
|
return $element->name() === 'description'; |
|
107
|
1 |
|
}) |
|
108
|
1 |
|
->first() |
|
109
|
1 |
|
->content(); |
|
110
|
|
|
$title = $elements |
|
111
|
1 |
|
->filter(static function(ElementInterface $element): bool { |
|
112
|
1 |
|
return $element->name() === 'title'; |
|
113
|
1 |
|
}) |
|
114
|
1 |
|
->first() |
|
115
|
1 |
|
->content(); |
|
116
|
|
|
$publicationDate = $elements |
|
117
|
1 |
|
->filter(static function(ElementInterface $element): bool { |
|
118
|
1 |
|
return $element->name() === 'pubDate'; |
|
119
|
1 |
|
}) |
|
120
|
1 |
|
->first() |
|
121
|
1 |
|
->content(); |
|
122
|
|
|
|
|
123
|
1 |
|
return $articles->add(Article::fetch( |
|
124
|
1 |
|
($this->make)(trim($link)), |
|
125
|
1 |
|
new Author(trim($author)), |
|
126
|
1 |
|
new Description(trim($description)), |
|
127
|
1 |
|
new Title(trim($title)), |
|
128
|
1 |
|
$this->clock->at(trim($publicationDate)) |
|
129
|
|
|
)); |
|
130
|
1 |
|
} |
|
131
|
|
|
); |
|
132
|
|
|
|
|
133
|
1 |
|
return $attributes->put( |
|
134
|
1 |
|
self::key(), |
|
135
|
1 |
|
new Attribute( |
|
136
|
1 |
|
self::key(), |
|
137
|
1 |
|
$articles |
|
138
|
|
|
) |
|
139
|
|
|
); |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
2 |
|
public static function key(): string |
|
143
|
|
|
{ |
|
144
|
2 |
|
return 'articles'; |
|
145
|
|
|
} |
|
146
|
|
|
} |
|
147
|
|
|
|
PersonalGalaxy /
RSS
