|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace eNTiDi\FeedReader; |
|
4
|
|
|
|
|
5
|
|
|
use GuzzleHttp\Client; |
|
6
|
|
|
use Psr\SimpleCache\CacheInterface; |
|
7
|
|
|
use SilverStripe\Core\Injector\Injector; |
|
8
|
|
|
use SilverStripe\ORM; |
|
9
|
|
|
use SilverStripe\ORM\ArrayList; |
|
10
|
|
|
use SilverStripe\ORM\FieldType\DBField; |
|
11
|
|
|
use SilverStripe\View\ArrayData; |
|
12
|
|
|
use SimpleXMLElement; |
|
13
|
|
|
|
|
14
|
|
|
class FeedReaderService |
|
15
|
|
|
{ |
|
16
|
|
|
private $url; |
|
17
|
|
|
private $expiration; |
|
18
|
|
|
private $options; |
|
19
|
|
|
private $summary_len; |
|
20
|
|
|
|
|
21
|
|
|
private static function getCache() |
|
22
|
|
|
{ |
|
23
|
|
|
return Injector::inst()->get(CacheInterface::class . '.FeedReader'); |
|
24
|
|
|
} |
|
25
|
|
|
|
|
26
|
|
|
private function getKey() |
|
27
|
|
|
{ |
|
28
|
|
|
return urlencode($this->url); |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
private static function dateObject($node) |
|
32
|
|
|
{ |
|
33
|
|
|
$text = (string) $node; |
|
34
|
|
|
if (empty($text)) { |
|
35
|
|
|
return null; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
$timestamp = strtotime($text); |
|
39
|
|
|
if ($timestamp === false) { |
|
40
|
|
|
return null; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
return DBField::create_field('DBDatetime', $timestamp); |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
private static function excerpt($html, $maxlen) |
|
47
|
|
|
{ |
|
48
|
|
|
// Strip HTML tags and convert blank chains to a single space |
|
49
|
|
|
$excerpt = trim(preg_replace('/\s+/', ' ', strip_tags($html))); |
|
50
|
|
|
if (strlen($excerpt) <= $maxlen) { |
|
51
|
|
|
return $excerpt; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
// Try to cut the excerpt on a word boundary |
|
55
|
|
|
$pivot = strrpos(substr($excerpt, 0, $maxlen - 2), ' '); |
|
56
|
|
|
if ($pivot === false || $pivot < $maxlen - 15) { |
|
57
|
|
|
$pivot = $maxlen - 3; |
|
58
|
|
|
} |
|
59
|
|
|
$excerpt = rtrim(substr($excerpt, 0, $pivot)); |
|
60
|
|
|
|
|
61
|
|
|
// Ellipsize the final result |
|
62
|
|
|
return rtrim($excerpt, '.') . '...'; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
private function appendRSS2Items(&$items, $xml) |
|
66
|
|
|
{ |
|
67
|
|
|
foreach ($xml->xpath('//channel/item') as $node) { |
|
68
|
|
|
$content = (string) $node->description; |
|
69
|
|
|
$row = new ArrayData([ |
|
70
|
|
|
'Id' => (string) $node->guid, |
|
71
|
|
|
'Link' => (string) $node->link, |
|
72
|
|
|
'Date' => self::dateObject($node->pubDate), |
|
73
|
|
|
'Title' => (string) $node->title, |
|
74
|
|
|
|
|
75
|
|
|
// RSS 2.0 does not have a summary field: generate it |
|
76
|
|
|
// from an excerpt of the "description" field |
|
77
|
|
|
'Summary' => self::excerpt($content, $this->getSummaryLen()), |
|
78
|
|
|
|
|
79
|
|
|
'Content' => $content |
|
80
|
|
|
]); |
|
81
|
|
|
$items->push($row); |
|
82
|
|
|
} |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
private function appendAtom1Items(&$items, $xml) |
|
86
|
|
|
{ |
|
87
|
|
|
$xml->registerXPathNamespace('A', 'http://www.w3.org/2005/Atom'); |
|
88
|
|
|
foreach ($xml->xpath('//A:feed/A:entry') as $node) { |
|
89
|
|
|
$summary = (string) $node->summary; |
|
90
|
|
|
if (! $node->content) { |
|
91
|
|
|
// Atom 1.0 does not require <content> elements, so |
|
92
|
|
|
// ensure it is at least populated with $summary |
|
93
|
|
|
$content = $summary; |
|
94
|
|
|
} elseif ($node->content->count() > 0) { |
|
95
|
|
|
$content = $node->content->children()[0]->asXML(); |
|
96
|
|
|
} else { |
|
97
|
|
|
$content = (string) $node->content; |
|
98
|
|
|
} |
|
99
|
|
|
$row = new ArrayData([ |
|
100
|
|
|
'Id' => (string) $node->id, |
|
101
|
|
|
'Link' => (string) $node->link['href'], |
|
102
|
|
|
'Date' => self::dateObject($node->updated), |
|
103
|
|
|
'Title' => (string) $node->title, |
|
104
|
|
|
'Summary' => $summary, |
|
105
|
|
|
'Content' => $content, |
|
106
|
|
|
]); |
|
107
|
|
|
$items->push($row); |
|
108
|
|
|
} |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
public function __construct($url, $expiration = 3600, $options = []) |
|
112
|
|
|
{ |
|
113
|
|
|
$this->url = $url; |
|
114
|
|
|
$this->expiration = $expiration; |
|
115
|
|
|
$this->options = $options; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
public function setSummaryLen($maxlen) |
|
119
|
|
|
{ |
|
120
|
|
|
$this->summary_len = $maxlen; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
public function getSummaryLen() |
|
124
|
|
|
{ |
|
125
|
|
|
return is_int($this->summary_len) ? $this->summary_len : 155; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
public function getItems() |
|
129
|
|
|
{ |
|
130
|
|
|
$cache = self::getCache(); |
|
131
|
|
|
$key = $this->getKey(); |
|
132
|
|
|
if ($cache->has($key)) { |
|
133
|
|
|
$items = $cache->get($key); |
|
134
|
|
|
} else { |
|
135
|
|
|
$client = new Client($this->options + [ 'timeout' => 2 ]); |
|
136
|
|
|
$response = $client->request('GET', $this->url); |
|
137
|
|
|
$code = $response->getStatusCode(); |
|
138
|
|
|
$data = $response->getBody()->getContents(); |
|
139
|
|
|
if ($code != 200) { |
|
140
|
|
|
user_error("RSS fetch error ($code). The response body is '$data'", E_USER_ERROR); |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
$xml = new SimpleXMLElement($data); |
|
144
|
|
|
$items = new ArrayList(); |
|
145
|
|
|
$this->appendRSS2Items($items, $xml); |
|
146
|
|
|
$this->appendAtom1Items($items, $xml); |
|
147
|
|
|
$cache->set($key, $items, $this->expiration); |
|
148
|
|
|
} |
|
149
|
|
|
return $items; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
public function clearCache() |
|
153
|
|
|
{ |
|
154
|
|
|
$cache = self::getCache(); |
|
155
|
|
|
$cache->delete($this->getKey()); |
|
156
|
|
|
} |
|
157
|
|
|
} |
|
158
|
|
|
|
ntd /
silverstripe-feedreader
