|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Colligator\Jobs; |
|
4
|
|
|
|
|
5
|
|
|
use Colligator\Collection; |
|
6
|
|
|
use Illuminate\Foundation\Bus\DispatchesJobs; |
|
7
|
|
|
use Log; |
|
8
|
|
|
use Phpoaipmh\Client; |
|
9
|
|
|
use Phpoaipmh\Endpoint; |
|
10
|
|
|
use Scriptotek\OaiPmh\ListRecordsResponse; |
|
11
|
|
|
use Storage; |
|
12
|
|
|
|
|
13
|
|
|
class OaiPmhHarvest extends Job |
|
14
|
|
|
{ |
|
15
|
|
|
use DispatchesJobs; |
|
16
|
|
|
|
|
17
|
|
|
public $name; |
|
18
|
|
|
public $url; |
|
19
|
|
|
public $schema; |
|
20
|
|
|
public $set; |
|
21
|
|
|
public $start; |
|
22
|
|
|
public $until; |
|
23
|
|
|
public $resume; |
|
24
|
|
|
public $maxRetries; |
|
25
|
|
|
public $sleepTimeOnError; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Start time for the full harvest. |
|
29
|
|
|
* |
|
30
|
|
|
* @var float |
|
31
|
|
|
*/ |
|
32
|
|
|
protected $startTime; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* Start time for the current batch. |
|
36
|
|
|
* |
|
37
|
|
|
* @var float |
|
38
|
|
|
*/ |
|
39
|
|
|
protected $batchTime; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Harvest position. |
|
43
|
|
|
* |
|
44
|
|
|
* @var int |
|
45
|
|
|
*/ |
|
46
|
|
|
protected $batchPos = 0; |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* @var Collection |
|
50
|
|
|
*/ |
|
51
|
|
|
public $collection; |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* Number of records retrieved between each emitted OaiPmhHarvestStatus event. |
|
55
|
|
|
* A too small number will cause CPU overhead. |
|
56
|
|
|
* |
|
57
|
|
|
* @var int |
|
58
|
|
|
*/ |
|
59
|
|
|
protected $statusUpdateEvery = 50; |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* Create a new job instance. |
|
63
|
|
|
* |
|
64
|
|
|
* @param string $name Harvest name from config |
|
65
|
|
|
* @param array $config Harvest config array (url, set, schema) |
|
66
|
|
|
* @param string $start Start date (optional) |
|
67
|
|
|
* @param string $until End date (optional) |
|
68
|
|
|
* @param string $resume Resumption token for continuing an aborted harvest (optional) |
|
69
|
|
|
*/ |
|
70
|
|
|
public function __construct($name, $config, $start = null, $until = null, $resume = null) |
|
71
|
|
|
{ |
|
72
|
|
|
$this->name = $name; |
|
73
|
|
|
$this->url = $config['url']; |
|
74
|
|
|
$this->schema = $config['schema']; |
|
75
|
|
|
$this->set = $config['set']; |
|
76
|
|
|
$this->start = $start; |
|
77
|
|
|
$this->until = $until; |
|
78
|
|
|
$this->resume = $resume; |
|
79
|
|
|
$this->maxRetries = array_get($config, 'max-retries', 1000); |
|
80
|
|
|
$this->sleepTimeOnError = array_get($config, 'sleep-time-on-error', 60); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
public function fromNetwork() |
|
84
|
|
|
{ |
|
85
|
|
|
$client = new Client($this->url); |
|
86
|
|
|
$endpoint = new Endpoint($client); |
|
87
|
|
|
|
|
88
|
|
|
$recordsHarvested = 0; |
|
89
|
|
|
|
|
90
|
|
|
// Loop over all records using an iterator that pulls in more data when |
|
91
|
|
|
// the buffer is exhausted. |
|
92
|
|
|
foreach ($endpoint->listRecords($this->schema, $this->start, $this->until, $this->set, $this->resume) as $record) { |
|
|
|
|
|
|
93
|
|
|
++$recordsHarvested; |
|
94
|
|
|
|
|
95
|
|
|
$this->dispatch(new ImportRecord($this->collection, $record->asXML())); |
|
96
|
|
|
|
|
97
|
|
|
if ($recordsHarvested % $this->statusUpdateEvery == 0) { |
|
98
|
|
|
$this->status($recordsHarvested, $recordsHarvested); |
|
99
|
|
|
} |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
return $recordsHarvested; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
/** |
|
106
|
|
|
* Execute the job. |
|
107
|
|
|
*/ |
|
108
|
|
|
public function handle() |
|
109
|
|
|
{ |
|
110
|
|
|
Log::info('[OaiPmhHarvest] Starting job. Requesting records from ' . ($this->start ?: '(no limit)') . ' until ' . ($this->until ?: '(no limit)') . '.'); |
|
111
|
|
|
|
|
112
|
|
|
// For timing |
|
113
|
|
|
$this->startTime = $this->batchTime = microtime(true) - 1; |
|
114
|
|
|
|
|
115
|
|
|
$this->collection = Collection::where('name', '=', $this->name)->first(); |
|
116
|
|
|
if (is_null($this->collection)) { |
|
117
|
|
|
$this->error("Collection '$this->name' not found in DB"); |
|
118
|
|
|
|
|
119
|
|
|
return; |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
$recordsHarvested = $this->fromNetwork(); |
|
123
|
|
|
|
|
124
|
|
|
Log::info('[OaiPmhHarvest] Harvest complete, got ' . $recordsHarvested . ' records.'); |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
/** |
|
128
|
|
|
* Output a status message. |
|
129
|
|
|
* |
|
130
|
|
|
* @param int $fetched |
|
131
|
|
|
* @param int $current |
|
132
|
|
|
*/ |
|
133
|
|
|
public function status($fetched, $current) |
|
134
|
|
|
{ |
|
135
|
|
|
$totalTime = microtime(true) - $this->startTime; |
|
136
|
|
|
$batchTime = microtime(true) - $this->batchTime; |
|
137
|
|
|
$mem = round(memory_get_usage() / 1024 / 102.4) / 10; |
|
138
|
|
|
|
|
139
|
|
|
$currentSpeed = ($fetched - $this->batchPos) / $batchTime; |
|
140
|
|
|
$avgSpeed = $fetched / $totalTime; |
|
141
|
|
|
|
|
142
|
|
|
$this->batchTime = microtime(true); |
|
143
|
|
|
$this->batchPos = $fetched; |
|
144
|
|
|
|
|
145
|
|
|
Log::debug(sprintf( |
|
146
|
|
|
'[OaiPmhHarvest] Got %d records so far - Recs/sec: %.1f (current), %.1f (avg) - Mem: %.1f MB.', |
|
147
|
|
|
$current, |
|
148
|
|
|
$currentSpeed, |
|
149
|
|
|
$avgSpeed, |
|
150
|
|
|
$mem |
|
151
|
|
|
)); |
|
152
|
|
|
} |
|
153
|
|
|
} |
|
154
|
|
|
|
scriptotek /
colligator-backend
Loading history...
If a method or function can return multiple different values and unless you are sure that you only can receive a single value in this context, we recommend to add an additional type check:
If this a common case that PHP Analyzer should handle natively, please let us know by opening an issue.