|
1
|
|
|
<?php namespace Scriptotek\OaiPmh; |
|
2
|
|
|
|
|
3
|
|
|
use GuzzleHttp\Client as HttpClient; |
|
4
|
|
|
use Evenement\EventEmitter; |
|
5
|
|
|
|
|
6
|
|
|
/** |
|
7
|
|
|
* OAI client |
|
8
|
|
|
*/ |
|
9
|
|
|
class Client extends EventEmitter |
|
10
|
|
|
{ |
|
11
|
|
|
|
|
12
|
|
|
// When we no longer need to support PHP 5.3: |
|
13
|
|
|
// - Upgrade to Evenement 2.0 and use trait instead of extending |
|
14
|
|
|
|
|
15
|
|
|
/** @var HttpClient */ |
|
16
|
|
|
protected $httpClient; |
|
17
|
|
|
|
|
18
|
|
|
/** @var string OAI service base URL */ |
|
19
|
|
|
public $url; |
|
20
|
|
|
|
|
21
|
|
|
/** @var string Requested schema for the returned records */ |
|
22
|
|
|
public $schema; |
|
23
|
|
|
|
|
24
|
|
|
/** @var string Some user agent string to identify our client */ |
|
25
|
|
|
public $userAgent; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var string|string[] Proxy configuration details. |
|
29
|
|
|
* |
|
30
|
|
|
* Either a string 'host:port' or an |
|
31
|
|
|
* array('host:port', 'username', 'password'). |
|
32
|
|
|
*/ |
|
33
|
|
|
public $proxy; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* @var string[] Array containing username and password |
|
37
|
|
|
*/ |
|
38
|
|
|
public $credentials; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* @var integer Max number of retries before giving up |
|
42
|
|
|
*/ |
|
43
|
|
|
public $maxRetries; |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* @var float Sleep time in seconds before retrying when getting an errorneous response |
|
47
|
|
|
*/ |
|
48
|
|
|
public $sleepTimeOnError; |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* @var integer Timeout for each request in seconds |
|
52
|
|
|
*/ |
|
53
|
|
|
public $timeout; |
|
54
|
|
|
|
|
55
|
|
|
protected function arrayGet($arr, $key, $default = null) |
|
56
|
|
|
{ |
|
57
|
|
|
return (isset($arr[$key])) ? $arr[$key] : $default; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
/** |
|
61
|
|
|
* Create a new client |
|
62
|
|
|
* |
|
63
|
|
|
* @param string $url Base URL to the OAI-PMH service |
|
64
|
|
|
* @param array $options Associative array of options |
|
65
|
|
|
* @param HttpClient $httpClient |
|
66
|
|
|
*/ |
|
67
|
|
|
public function __construct($url, $options = array(), $httpClient = null) |
|
68
|
|
|
{ |
|
69
|
|
|
$this->url = $url; |
|
70
|
|
|
$this->httpClient = $httpClient ?: new HttpClient; |
|
71
|
|
|
|
|
72
|
|
|
$this->schema = $this->arrayGet($options, 'schema', 'marcxchange'); |
|
73
|
|
|
$this->userAgent = $this->arrayGet($options, 'user-agent', 'php-oaipmh-client'); |
|
74
|
|
|
$this->credentials = $this->arrayGet($options, 'credentials'); |
|
75
|
|
|
$this->proxy = $this->arrayGet($options, 'proxy'); |
|
76
|
|
|
$this->maxRetries = $this->arrayGet($options, 'max-retries', 30); |
|
77
|
|
|
$this->sleepTimeOnError = $this->arrayGet($options, 'sleep-time-on-error', 1.0); |
|
78
|
|
|
$this->timeout = $this->arrayGet($options, 'timeout', 60.0); |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
/** |
|
82
|
|
|
* Get HTTP client configuration options (authentication, proxy) |
|
83
|
|
|
* |
|
84
|
|
|
* @return array |
|
85
|
|
|
*/ |
|
86
|
|
|
public function getHttpOptions() |
|
87
|
|
|
{ |
|
88
|
|
|
$options = array( |
|
89
|
|
|
'headers' => $this->getHttpHeaders(), |
|
90
|
|
|
'connect_timeout' => $this->timeout, |
|
91
|
|
|
'timeout' => $this->timeout, |
|
92
|
|
|
); |
|
93
|
|
|
if ($this->credentials) { |
|
|
|
|
|
|
94
|
|
|
$options['auth'] = $this->credentials; |
|
95
|
|
|
} |
|
96
|
|
|
if ($this->proxy) { |
|
97
|
|
|
$options['proxy'] = $this->proxy; |
|
98
|
|
|
} |
|
99
|
|
|
return $options; |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* Get HTTP client headers |
|
104
|
|
|
* |
|
105
|
|
|
* @return array |
|
106
|
|
|
*/ |
|
107
|
|
|
public function getHttpHeaders() |
|
108
|
|
|
{ |
|
109
|
|
|
$headers = array( |
|
110
|
|
|
'Accept' => 'application/xml' |
|
111
|
|
|
); |
|
112
|
|
|
if ($this->userAgent) { |
|
113
|
|
|
$headers['User-Agent'] = $this->userAgent; |
|
114
|
|
|
} |
|
115
|
|
|
return $headers; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* Construct the URL for an OAI-PMH query |
|
120
|
|
|
* |
|
121
|
|
|
* @param string $verb The OAI-PMH verb |
|
122
|
|
|
* @param array $arguments OAI-PMH arguments |
|
123
|
|
|
* @return string |
|
124
|
|
|
*/ |
|
125
|
|
|
public function urlBuilder($verb, $arguments = array()) |
|
126
|
|
|
{ |
|
127
|
|
|
$qs = array( |
|
128
|
|
|
'verb' => $verb, |
|
129
|
|
|
'metadataPrefix' => $this->schema, |
|
130
|
|
|
); |
|
131
|
|
|
|
|
132
|
|
|
foreach ($arguments as $key => $value) { |
|
133
|
|
|
$qs[$key] = $value; |
|
134
|
|
|
if (is_null($value)) { |
|
135
|
|
|
// Allow removal of default arguments like 'metadataPrefix' |
|
136
|
|
|
unset($qs[$key]); |
|
137
|
|
|
} |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
return $this->url . '?' . http_build_query($qs); |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
/** |
|
144
|
|
|
* Perform a single OAI-PMH request |
|
145
|
|
|
* |
|
146
|
|
|
* @param string $verb The OAI-PMH verb |
|
147
|
|
|
* @param array $arguments OAI-PMH arguments |
|
148
|
|
|
* @return string |
|
149
|
|
|
* @throws ConnectionError |
|
150
|
|
|
*/ |
|
151
|
|
|
public function request($verb, $arguments) |
|
152
|
|
|
{ |
|
153
|
|
|
$this->emit('request.start', array( |
|
154
|
|
|
'verb' => $verb, |
|
155
|
|
|
'arguments' => $arguments |
|
156
|
|
|
)); |
|
157
|
|
|
$url = $this->urlBuilder($verb, $arguments); |
|
158
|
|
|
$attempt = 0; |
|
159
|
|
|
while (true) { |
|
160
|
|
|
try { |
|
161
|
|
|
$res = $this->httpClient->get($url, $this->getHttpOptions()); |
|
162
|
|
|
break; |
|
163
|
|
|
} catch (\GuzzleHttp\Exception\ConnectException $e) { |
|
164
|
|
|
// Thrown in case of a networking error (connection timeout, DNS errors, etc.) |
|
165
|
|
|
$this->emit('request.error', array( |
|
166
|
|
|
'message' => $e->getMessage(), |
|
167
|
|
|
)); |
|
168
|
|
|
time_nanosleep(intval($this->sleepTimeOnError), intval($this->sleepTimeOnError * 1000000000)); |
|
169
|
|
|
} catch (\GuzzleHttp\Exception\ServerException $e) { |
|
170
|
|
|
// Thrown in case of 500 errors |
|
171
|
|
|
$this->emit('request.error', array( |
|
172
|
|
|
'message' => $e->getMessage(), |
|
173
|
|
|
)); |
|
174
|
|
|
time_nanosleep(intval($this->sleepTimeOnError), intval($this->sleepTimeOnError * 1000000000)); |
|
175
|
|
|
} |
|
176
|
|
|
$attempt++; |
|
177
|
|
|
if ($attempt > $this->maxRetries) { |
|
178
|
|
|
throw new ConnectionError('Failed to get a response from the server. Max retries (' . $this->maxRetries . ') exceeded.'); |
|
179
|
|
|
} |
|
180
|
|
|
} |
|
181
|
|
|
$body = (string) $res->getBody(); |
|
|
|
|
|
|
182
|
|
|
$this->emit('request.complete', array( |
|
183
|
|
|
'verb' => $verb, |
|
184
|
|
|
'arguments' => $arguments, |
|
185
|
|
|
'response' => $body |
|
186
|
|
|
)); |
|
187
|
|
|
return $body; |
|
188
|
|
|
} |
|
189
|
|
|
|
|
190
|
|
|
/** |
|
191
|
|
|
* Perform a GetRecord request |
|
192
|
|
|
* |
|
193
|
|
|
* @param string $identifier |
|
194
|
|
|
* @return GetRecordResponse |
|
195
|
|
|
*/ |
|
196
|
|
|
public function record($identifier) |
|
197
|
|
|
{ |
|
198
|
|
|
$data = $this->request('GetRecord', array('identifier' => $identifier)); |
|
199
|
|
|
return new GetRecordResponse($data); |
|
200
|
|
|
} |
|
201
|
|
|
|
|
202
|
|
|
/** |
|
203
|
|
|
* Perform a ListRecords request and return an iterator over the records |
|
204
|
|
|
* |
|
205
|
|
|
* @param string $from Start date |
|
206
|
|
|
* @param string $until End date |
|
207
|
|
|
* @param string $set Data set |
|
208
|
|
|
* @param string $resumptionToken To resume a harvest |
|
209
|
|
|
* @param array $extraParams Extra GET parameters |
|
210
|
|
|
* @return Records |
|
211
|
|
|
*/ |
|
212
|
|
|
public function records($from, $until, $set, $resumptionToken = null, $extraParams = array()) |
|
213
|
|
|
{ |
|
214
|
|
|
return new Records($from, $until, $set, $this, $resumptionToken, $extraParams); |
|
|
|
|
|
|
215
|
|
|
} |
|
216
|
|
|
} |
|
217
|
|
|
|
This check marks implicit conversions of arrays to boolean values in a comparison. While in PHP an empty array is considered to be equal (but not identical) to false, this is not always apparent.
Consider making the comparison explicit by using
empty(..)or! empty(...)instead.