|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace LegalThings\DataEnricher\Processor; |
|
4
|
|
|
|
|
5
|
|
|
use LegalThings\DataEnricher\Node; |
|
6
|
|
|
use LegalThings\DataEnricher\Processor; |
|
7
|
|
|
use GuzzleHttp\Client; |
|
8
|
|
|
use GuzzleHttp\Psr7\Response; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Load an external source |
|
12
|
|
|
*/ |
|
13
|
|
|
class Http implements Processor |
|
14
|
|
|
{ |
|
15
|
|
|
use Processor\Implementation; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* Get the URL from a node |
|
19
|
|
|
* |
|
20
|
|
|
* @param Node $node |
|
21
|
|
|
* @return string |
|
22
|
|
|
*/ |
|
23
|
|
|
protected function getUrl($node) |
|
24
|
|
|
{ |
|
25
|
|
|
$url = $node->getInstruction($this); |
|
26
|
|
|
|
|
27
|
|
|
$type = (is_object($url) ? get_class($url) . ' ' : '') . gettype($url); |
|
28
|
|
|
if (!assert(is_string($url), "Expected '{$this->property}' to be a string, but got a $type")) { |
|
29
|
|
|
return null; |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
return $url; |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Do async request for node |
|
37
|
|
|
* |
|
38
|
|
|
* @param Node $node |
|
39
|
|
|
*/ |
|
40
|
|
|
public function applyToNode(Node $node) |
|
41
|
|
|
{ |
|
42
|
|
|
$url = $this->getUrl($node); |
|
43
|
|
|
if (!isset($url)) { |
|
44
|
|
|
return; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
$client = new Client(['http_errors' => false]); |
|
48
|
|
|
$promise = $client->getAsync($url)->then(function (Response $response) use ($node) { |
|
49
|
|
|
$this->applyResult($node, $response); |
|
50
|
|
|
}); |
|
51
|
|
|
|
|
52
|
|
|
$node->setResult($promise); |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* Apply results to a node |
|
57
|
|
|
* |
|
58
|
|
|
* @param Node $node |
|
59
|
|
|
* @param Response $response |
|
60
|
|
|
*/ |
|
61
|
|
|
protected function applyResult(Node $node, Response $response) |
|
62
|
|
|
{ |
|
63
|
|
|
$result = null; |
|
64
|
|
|
|
|
65
|
|
|
if ($this->hasExpectedResponse($node, $response)) { |
|
66
|
|
|
$result = json_decode($response->getBody()); |
|
67
|
|
|
|
|
68
|
|
|
if (!$result) { |
|
69
|
|
|
$url = $this->getUrl($node); |
|
70
|
|
|
trigger_error("Failed to fetch '$url': Corrupt JSON response", E_USER_WARNING); |
|
71
|
|
|
} |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
$node->setResult($result); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
/** |
|
78
|
|
|
* Check if we got an expected response |
|
79
|
|
|
* |
|
80
|
|
|
* @param Node $node |
|
81
|
|
|
* @param Response $response |
|
82
|
|
|
* @return boolean |
|
83
|
|
|
*/ |
|
84
|
|
|
protected function hasExpectedResponse(Node $node, Response $response) |
|
85
|
|
|
{ |
|
86
|
|
|
$status = $response->getStatusCode(); |
|
87
|
|
|
$contentType = preg_replace('/\s*;.*$/', '', $response->getHeaderLine('Content-Type')); |
|
88
|
|
|
|
|
89
|
|
|
if ($status == 404) { |
|
90
|
|
|
return false; |
|
91
|
|
|
} elseif ($status >= 300 || !in_array($contentType, ['application/json', 'text/plain'])) { |
|
92
|
|
|
$url = $this->getUrl($node); |
|
93
|
|
|
|
|
94
|
|
|
if ($contentType === 'text/plain') { |
|
95
|
|
|
$message = $response->getBody(); |
|
96
|
|
|
} else { |
|
97
|
|
|
$message = "Server responded with a $status status and $contentType"; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
trigger_error("Failed to fetch '$url': $message", E_USER_WARNING); |
|
101
|
|
|
return false; |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
return true; |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
legalthings /
data-enricher
