|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Fusonic\OpenGraph; |
|
4
|
|
|
|
|
5
|
|
|
use DOMElement; |
|
6
|
|
|
use Fusonic\Linq\Linq; |
|
7
|
|
|
use Fusonic\OpenGraph\Objects\ObjectBase; |
|
8
|
|
|
use Fusonic\OpenGraph\Objects\Website; |
|
9
|
|
|
use LogicException; |
|
10
|
|
|
use Psr\Http\Client\ClientExceptionInterface; |
|
11
|
|
|
use Psr\Http\Client\ClientInterface; |
|
12
|
|
|
use Psr\Http\Message\RequestFactoryInterface; |
|
13
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* Consumer that extracts Open Graph data from either a URL or a HTML string. |
|
17
|
|
|
*/ |
|
18
|
|
|
class Consumer |
|
19
|
|
|
{ |
|
20
|
|
|
private ?ClientInterface $client; |
|
|
|
|
|
|
21
|
|
|
private ?RequestFactoryInterface $requestFactory; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* When enabled, crawler will read content of title and meta description if no |
|
25
|
|
|
* Open Graph data is provided by target page. |
|
26
|
|
|
*/ |
|
27
|
|
|
public bool $useFallbackMode = false; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* When enabled, crawler will throw exceptions for some crawling errors like unexpected |
|
31
|
|
|
* Open Graph elements. |
|
32
|
|
|
*/ |
|
33
|
|
|
public bool $debug = false; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* @param ClientInterface|null $client A PSR-18 ClientInterface implementation. |
|
37
|
|
|
* @param RequestFactoryInterface|null $requestFactory A PSR-17 RequestFactoryInterface implementation. |
|
38
|
|
|
*/ |
|
39
|
16 |
|
public function __construct(?ClientInterface $client = null, ?RequestFactoryInterface $requestFactory = null) |
|
40
|
|
|
{ |
|
41
|
16 |
|
$this->client = $client; |
|
42
|
16 |
|
$this->requestFactory = $requestFactory; |
|
43
|
16 |
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* Fetches HTML content from the given URL and then crawls it for Open Graph data. |
|
47
|
|
|
* |
|
48
|
|
|
* @param string $url URL to be crawled. |
|
49
|
|
|
* |
|
50
|
|
|
* @return ObjectBase |
|
51
|
|
|
* |
|
52
|
|
|
* @throws ClientExceptionInterface |
|
53
|
|
|
*/ |
|
54
|
|
|
public function loadUrl(string $url): ObjectBase |
|
55
|
|
|
{ |
|
56
|
|
|
if ($this->client === null) { |
|
57
|
|
|
throw new LogicException( |
|
58
|
|
|
"To use loadUrl() you must provide \$client and \$requestFactory when instantiating the consumer." |
|
59
|
|
|
); |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
$request = $this->requestFactory->createRequest("GET", $url); |
|
63
|
|
|
$response = $this->client->sendRequest($request); |
|
64
|
|
|
|
|
65
|
|
|
return $this->loadHtml($response->getBody()->getContents(), $url); |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
/** |
|
69
|
|
|
* Crawls the given HTML string for OpenGraph data. |
|
70
|
|
|
* |
|
71
|
|
|
* @param string $html HTML string, usually whole content of crawled web resource. |
|
72
|
|
|
* @param string $fallbackUrl URL to use when fallback mode is enabled. |
|
73
|
|
|
* |
|
74
|
|
|
* @return ObjectBase |
|
75
|
|
|
*/ |
|
76
|
16 |
|
public function loadHtml(string $html, string $fallbackUrl = null): ObjectBase |
|
77
|
|
|
{ |
|
78
|
|
|
// Extract all data that can be found |
|
79
|
16 |
|
$page = $this->extractOpenGraphData($html); |
|
80
|
|
|
|
|
81
|
|
|
// Use the user's URL as fallback |
|
82
|
13 |
|
if ($this->useFallbackMode && $page->url === null) { |
|
83
|
1 |
|
$page->url = $fallbackUrl; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
// Return result |
|
87
|
13 |
|
return $page; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
16 |
|
private function extractOpenGraphData(string $content): ObjectBase |
|
91
|
|
|
{ |
|
92
|
16 |
|
$crawler = new Crawler; |
|
93
|
16 |
|
$crawler->addHTMLContent($content, 'UTF-8'); |
|
94
|
|
|
|
|
95
|
16 |
|
$properties = []; |
|
96
|
16 |
|
foreach(['name', 'property'] as $t) |
|
97
|
|
|
{ |
|
98
|
|
|
// Get all meta-tags starting with "og:" |
|
99
|
16 |
|
$ogMetaTags = $crawler->filter("meta[{$t}^='og:']"); |
|
100
|
|
|
// Create clean property array |
|
101
|
16 |
|
$props = Linq::from($ogMetaTags) |
|
102
|
16 |
|
->select( |
|
103
|
|
|
function (DOMElement $tag) use ($t) { |
|
104
|
13 |
|
$name = strtolower(trim($tag->getAttribute($t))); |
|
105
|
13 |
|
$value = trim($tag->getAttribute("content")); |
|
106
|
13 |
|
return new Property($name, $value); |
|
107
|
16 |
|
} |
|
108
|
|
|
) |
|
109
|
16 |
|
->toArray(); |
|
110
|
16 |
|
$properties = array_merge($properties, $props); |
|
111
|
|
|
|
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
// Create new object of the correct type |
|
115
|
16 |
|
$typeProperty = Linq::from($properties) |
|
116
|
16 |
|
->firstOrNull( |
|
117
|
|
|
function (Property $property) { |
|
118
|
13 |
|
return $property->key === Property::TYPE; |
|
119
|
16 |
|
} |
|
120
|
|
|
); |
|
121
|
16 |
|
switch ($typeProperty !== null ? $typeProperty->value : null) { |
|
122
|
|
|
default: |
|
123
|
16 |
|
$object = new Website(); |
|
124
|
16 |
|
break; |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
// Assign all properties to the object |
|
128
|
16 |
|
$object->assignProperties($properties, $this->debug); |
|
129
|
|
|
|
|
130
|
|
|
// Fallback for url |
|
131
|
13 |
|
if ($this->useFallbackMode && !$object->url) { |
|
132
|
2 |
|
$urlElement = $crawler->filter("link[rel='canonical']")->first(); |
|
133
|
2 |
|
if ($urlElement->count() > 0) { |
|
134
|
1 |
|
$object->url = trim($urlElement->attr("href")); |
|
135
|
|
|
} |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
// Fallback for title |
|
139
|
13 |
|
if ($this->useFallbackMode && !$object->title) { |
|
140
|
2 |
|
$titleElement = $crawler->filter("title")->first(); |
|
141
|
2 |
|
if ($titleElement->count() > 0) { |
|
142
|
2 |
|
$object->title = trim($titleElement->text()); |
|
143
|
|
|
} |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
|
|
// Fallback for description |
|
147
|
13 |
|
if ($this->useFallbackMode && !$object->description) { |
|
148
|
2 |
|
$descriptionElement = $crawler->filter("meta[property='description']")->first(); |
|
149
|
2 |
|
if ($descriptionElement->count() > 0) { |
|
150
|
2 |
|
$object->description = trim($descriptionElement->attr("content")); |
|
151
|
|
|
} |
|
152
|
|
|
} |
|
153
|
|
|
|
|
154
|
13 |
|
return $object; |
|
155
|
|
|
} |
|
156
|
|
|
} |
|
157
|
|
|
|
fusonic /
opengraph
Loading history...