|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Mpyw\OpenGraph; |
|
4
|
|
|
|
|
5
|
|
|
use DOMElement; |
|
6
|
|
|
use Mpyw\OpenGraph\Objects\ObjectBase; |
|
7
|
|
|
use Mpyw\OpenGraph\Objects\Website; |
|
8
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Consumer that extracts Open Graph data from either a URL or a HTML string. |
|
12
|
|
|
*/ |
|
13
|
|
|
class Consumer |
|
14
|
|
|
{ |
|
15
|
|
|
use GenericHelper; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* When enabled, crawler will read content of title and meta description if no |
|
19
|
|
|
* Open Graph data is provided by target page. |
|
20
|
|
|
* |
|
21
|
|
|
* @var bool |
|
22
|
|
|
*/ |
|
23
|
|
|
public $useFallbackMode = false; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* When enabled, crawler will throw exceptions for some crawling errors like unexpected |
|
27
|
|
|
* Open Graph elements. |
|
28
|
|
|
* |
|
29
|
|
|
* @var bool |
|
30
|
|
|
*/ |
|
31
|
|
|
public $debug = false; |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @param string $html HTML string, usually whole content of crawled web resource. |
|
35
|
|
|
* @param null|string $fallbackUrl URL to use when fallback mode is enabled. |
|
36
|
|
|
* @return ObjectBase |
|
37
|
|
|
*/ |
|
38
|
17 |
|
public function loadHtml(string $html, ?string $fallbackUrl = null) |
|
39
|
|
|
{ |
|
40
|
|
|
// Note: Avoid Crawler::evaluate() throwing LogicException against empty HTML |
|
41
|
17 |
|
$crawler = $this->newCrawler($this->emptyStringAsNull(trim($html)) ?? '<!DOCTYPE html><html></html>'); |
|
42
|
|
|
|
|
43
|
17 |
|
$properties = $this->newProperties($crawler); |
|
44
|
|
|
|
|
45
|
|
|
$object = $this |
|
46
|
17 |
|
->newObject($crawler) |
|
47
|
17 |
|
->assignProperties($properties, $this->debug); |
|
48
|
|
|
|
|
49
|
14 |
|
if ($this->useFallbackMode) { |
|
50
|
2 |
|
$this->fallback($object, $crawler, $fallbackUrl); |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
14 |
|
return $object; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* @param Crawler $crawler |
|
58
|
|
|
* @return Property[] |
|
59
|
|
|
*/ |
|
60
|
17 |
|
protected function newProperties(Crawler $crawler): array |
|
61
|
|
|
{ |
|
62
|
17 |
|
return array_map( |
|
63
|
17 |
|
[$this, 'newProperty'], |
|
64
|
17 |
|
iterator_to_array($crawler->filterXPath('//meta[starts-with(@property, "og:") or starts-with(@name, "og:")]'), false) |
|
65
|
|
|
); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* @param string $content |
|
70
|
|
|
* @return Crawler |
|
71
|
|
|
*/ |
|
72
|
17 |
|
protected function newCrawler(string $content) |
|
73
|
|
|
{ |
|
74
|
17 |
|
$crawler = new Crawler(); |
|
75
|
17 |
|
$crawler->addHTMLContent($content); |
|
76
|
17 |
|
return $crawler; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @param DOMElement $tag |
|
81
|
|
|
* @return Property |
|
82
|
|
|
*/ |
|
83
|
13 |
|
protected function newProperty(DOMElement $tag) |
|
84
|
|
|
{ |
|
85
|
13 |
|
$name = trim($tag->getAttribute('name') ?: $tag->getAttribute('property')); |
|
86
|
13 |
|
$value = trim($tag->getAttribute('content')); |
|
87
|
13 |
|
return new Property($name, $value); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* @param Crawler $crawler |
|
92
|
|
|
* @return ObjectBase |
|
93
|
|
|
*/ |
|
94
|
17 |
|
protected function newObject(Crawler $crawler) |
|
95
|
|
|
{ |
|
96
|
17 |
|
switch ($crawler->evaluate('normalize-space(//meta[@property="og:type" or @name="og:type"]/@content)')[0] ?? null) { |
|
97
|
|
|
default: |
|
98
|
17 |
|
return new Website(); |
|
99
|
|
|
} |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* @param ObjectBase $object |
|
104
|
|
|
* @param Crawler $crawler |
|
105
|
|
|
* @param null|string $fallbackUrl |
|
106
|
|
|
* @return $this |
|
107
|
|
|
*/ |
|
108
|
2 |
|
protected function fallback(ObjectBase $object, Crawler $crawler, ?string $fallbackUrl = null) |
|
109
|
|
|
{ |
|
110
|
|
|
return $this |
|
111
|
2 |
|
->fallbackForUrl($object, $crawler, $fallbackUrl) |
|
112
|
2 |
|
->fallbackForTitle($object, $crawler) |
|
113
|
2 |
|
->fallbackForDescription($object, $crawler); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* @param ObjectBase $object |
|
118
|
|
|
* @param Crawler $crawler |
|
119
|
|
|
* @param null|string $fallbackUrl |
|
120
|
|
|
* @return $this |
|
121
|
|
|
*/ |
|
122
|
2 |
|
protected function fallbackForUrl(ObjectBase $object, Crawler $crawler, ?string $fallbackUrl = null) |
|
123
|
|
|
{ |
|
124
|
2 |
|
$object->url = $object->url |
|
125
|
2 |
|
?? $this->emptyStringAsNull($crawler->evaluate('normalize-space(//link[@rel="canonical"]/@href)')[0]) |
|
126
|
1 |
|
?? $this->emptyStringAsNull($fallbackUrl) |
|
127
|
2 |
|
?? $object->url; |
|
128
|
|
|
|
|
129
|
2 |
|
return $this; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
/** |
|
133
|
|
|
* @param ObjectBase $object |
|
134
|
|
|
* @param Crawler $crawler |
|
135
|
|
|
* @return $this |
|
136
|
|
|
*/ |
|
137
|
2 |
|
protected function fallbackForTitle(ObjectBase $object, Crawler $crawler) |
|
138
|
|
|
{ |
|
139
|
2 |
|
$object->title = $object->title |
|
140
|
2 |
|
?? $this->emptyStringAsNull($crawler->evaluate('normalize-space(//title)')[0]) |
|
141
|
|
|
?? $this->emptyStringAsNull($crawler->evaluate('normalize-space(//h1)')[0]) |
|
142
|
|
|
?? $this->emptyStringAsNull($crawler->evaluate('normalize-space(//h2)')[0]) |
|
143
|
2 |
|
?? $object->title; |
|
144
|
|
|
|
|
145
|
2 |
|
return $this; |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
|
|
/** |
|
149
|
|
|
* @param ObjectBase $object |
|
150
|
|
|
* @param Crawler $crawler |
|
151
|
|
|
* @return $this |
|
152
|
|
|
*/ |
|
153
|
2 |
|
protected function fallbackForDescription(ObjectBase $object, Crawler $crawler) |
|
154
|
|
|
{ |
|
155
|
2 |
|
$object->description = $object->description |
|
156
|
2 |
|
?? $this->emptyStringAsNull($crawler->evaluate('normalize-space(//meta[@property="description" or @name="description"]/@content)')[0]) |
|
157
|
|
|
?? $this->emptyStringAsNull($crawler->evaluate('normalize-space(//p)')[0]) |
|
158
|
2 |
|
?? $object->description; |
|
159
|
|
|
|
|
160
|
2 |
|
return $this; |
|
161
|
|
|
} |
|
162
|
|
|
} |
|
163
|
|
|
|
mpyw /
opengraph
