|
1
|
|
|
<?php
|
|
2
|
|
|
|
|
3
|
|
|
namespace Radowoj\Searcher\SearchProvider;
|
|
4
|
|
|
|
|
5
|
|
|
use stdClass;
|
|
6
|
|
|
|
|
7
|
|
|
use GuzzleHttp\Client as GuzzleClient;
|
|
8
|
|
|
use Psr\Http\Message\ResponseInterface as Psr7Response;
|
|
9
|
|
|
|
|
10
|
|
|
use Radowoj\Searcher\SearchResult\Collection;
|
|
11
|
|
|
use Radowoj\Searcher\SearchResult\ICollection;
|
|
12
|
|
|
use Radowoj\Searcher\SearchResult\Item;
|
|
13
|
|
|
use Radowoj\Searcher\SearchResult\IItem;
|
|
14
|
|
|
|
|
15
|
|
|
use Radowoj\Searcher\Exceptions\Exception;
|
|
16
|
|
|
use Radowoj\Searcher\Exceptions\QuotaExceededException;
|
|
17
|
|
|
use Radowoj\Searcher\Exceptions\RateLimitExceededException;
|
|
18
|
|
|
|
|
19
|
|
|
class Bing extends SearchProvider implements ISearchProvider
|
|
20
|
|
|
{
|
|
21
|
|
|
const URI = 'https://api.cognitive.microsoft.com/bing/v5.0/search?';
|
|
22
|
|
|
const API_KEY_HEADER = 'Ocp-Apim-Subscription-Key';
|
|
23
|
|
|
|
|
24
|
|
|
protected $apiKey = null;
|
|
25
|
|
|
|
|
26
|
|
|
protected $guzzle = null;
|
|
27
|
|
|
|
|
28
|
10 |
|
public function __construct(GuzzleClient $guzzle, string $apiKey)
|
|
29
|
|
|
{
|
|
30
|
10 |
|
$this->apiKey = $apiKey;
|
|
31
|
10 |
|
$this->guzzle = $guzzle;
|
|
32
|
10 |
|
}
|
|
33
|
|
|
|
|
34
|
|
|
|
|
35
|
9 |
|
protected function searchRequest(string $query, int $limit, int $offset) : stdClass
|
|
36
|
|
|
{
|
|
37
|
|
|
$params = [
|
|
38
|
9 |
|
'q' => $query,
|
|
39
|
9 |
|
'count' => $limit,
|
|
40
|
9 |
|
'offset' => $offset
|
|
41
|
|
|
];
|
|
42
|
|
|
|
|
43
|
9 |
|
$queryString = http_build_query($params);
|
|
44
|
|
|
|
|
45
|
9 |
|
$uri = self::URI . $queryString;
|
|
46
|
|
|
|
|
47
|
9 |
|
$result = $this->guzzle->request(
|
|
48
|
9 |
|
'GET',
|
|
49
|
|
|
$uri, [
|
|
50
|
|
|
'headers' => [
|
|
51
|
9 |
|
self::API_KEY_HEADER => $this->apiKey,
|
|
52
|
9 |
|
],
|
|
53
|
|
|
'http_errors' => false,
|
|
54
|
|
|
]
|
|
55
|
|
|
);
|
|
56
|
|
|
|
|
57
|
9 |
|
return $this->decodeResponse($result);
|
|
58
|
|
|
}
|
|
59
|
|
|
|
|
60
|
|
|
|
|
61
|
|
|
/**
|
|
62
|
|
|
* Handle response based on HTTP status code (catches 400s - usually quota or rate limit,
|
|
63
|
|
|
* so authorisation errors and other stuff will be thrown as generic Searcher exception)
|
|
64
|
|
|
*
|
|
65
|
|
|
* On status == 200 it simply returns json-decoded response.
|
|
66
|
|
|
*
|
|
67
|
|
|
* @param Psr7Response $result result from Guzzle
|
|
68
|
|
|
* @return array
|
|
69
|
|
|
*/
|
|
70
|
9 |
|
protected function decodeResponse(Psr7Response $result) : stdClass
|
|
71
|
|
|
{
|
|
72
|
9 |
|
switch($result->getStatusCode()) {
|
|
73
|
9 |
|
case 200:
|
|
74
|
6 |
|
return json_decode($result->getBody());
|
|
75
|
3 |
|
case 403: //Out of call volume quota
|
|
76
|
1 |
|
throw new QuotaExceededException($result->getReasonPhrase());
|
|
77
|
2 |
|
case 429: //Rate limit is exceeded
|
|
78
|
1 |
|
throw new RateLimitExceededException($result->getReasonPhrase());
|
|
79
|
|
|
default:
|
|
80
|
1 |
|
throw new Exception("Bing API responded with HTTP status {$result->getStatusCode()} - {$result->getReasonPhrase()}");
|
|
81
|
|
|
}
|
|
82
|
|
|
}
|
|
83
|
|
|
|
|
84
|
|
|
|
|
85
|
6 |
|
protected function validateRequestResult(stdClass $result)
|
|
86
|
|
|
{
|
|
87
|
6 |
|
if (!isset($result->_type) || $result->_type !== 'SearchResponse') {
|
|
88
|
1 |
|
throw new Exception("Invalid Bing API response: " . print_r($result, 1));
|
|
89
|
|
|
}
|
|
90
|
5 |
|
}
|
|
91
|
|
|
|
|
92
|
|
|
|
|
93
|
5 |
|
protected function enforceLimit(stdClass $result, int $limit) : stdClass
|
|
94
|
|
|
{
|
|
95
|
5 |
|
if (!isset($result->webPages->value)) {
|
|
96
|
1 |
|
return $result;
|
|
97
|
|
|
}
|
|
98
|
4 |
|
$result->webPages->value = array_slice($result->webPages->value, 0, $limit);
|
|
99
|
4 |
|
return $result;
|
|
100
|
|
|
}
|
|
101
|
|
|
|
|
102
|
|
|
|
|
103
|
5 |
|
protected function extractResults(stdClass $result) : array
|
|
104
|
|
|
{
|
|
105
|
5 |
|
return isset($result->webPages->value)
|
|
106
|
4 |
|
? $result->webPages->value
|
|
107
|
5 |
|
: [];
|
|
108
|
|
|
}
|
|
109
|
|
|
|
|
110
|
|
|
|
|
111
|
5 |
|
protected function extractTotalMatches(stdClass $result) : int
|
|
112
|
|
|
{
|
|
113
|
5 |
|
return isset($result->webPages->totalEstimatedMatches)
|
|
114
|
4 |
|
? $result->webPages->totalEstimatedMatches
|
|
115
|
5 |
|
: 0;
|
|
116
|
|
|
}
|
|
117
|
|
|
|
|
118
|
|
|
|
|
119
|
1 |
|
protected function populateItem(stdClass $item) : IItem
|
|
120
|
|
|
{
|
|
121
|
1 |
|
return new Item([
|
|
122
|
1 |
|
'url' => preg_match('/^https?:\/\//', $item->url)
|
|
123
|
1 |
|
? $item->url
|
|
124
|
1 |
|
: "http://{$item->url}",
|
|
125
|
1 |
|
'title' => $item->name,
|
|
126
|
1 |
|
'description' => $item->snippet,
|
|
127
|
|
|
]);
|
|
128
|
|
|
}
|
|
129
|
|
|
|
|
130
|
|
|
}
|
|
131
|
|
|
|
radowoj /
searcher
