1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace SilverStripe\TextExtraction\Extractor; |
4
|
|
|
|
5
|
|
|
use SilverStripe\Assets\File; |
6
|
|
|
use SilverStripe\Core\Environment; |
7
|
|
|
use SilverStripe\Core\Injector\Injector; |
8
|
|
|
use SilverStripe\TextExtraction\Rest\TikaRestClient; |
9
|
|
|
|
10
|
|
|
/** |
11
|
|
|
* Enables text extraction of file content via the Tika Rest Server |
12
|
|
|
* |
13
|
|
|
* {@link http://tika.apache.org/1.7/gettingstarted.html} |
14
|
|
|
*/ |
15
|
|
|
class TikaServerTextExtractor extends FileTextExtractor |
16
|
|
|
{ |
17
|
|
|
/** |
18
|
|
|
* Tika server is pretty efficient so use it immediately if available |
19
|
|
|
* |
20
|
|
|
* @var integer |
21
|
|
|
* @config |
22
|
|
|
*/ |
23
|
|
|
private static $priority = 80; |
|
|
|
|
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* Server endpoint |
27
|
|
|
* |
28
|
|
|
* @var string |
29
|
|
|
* @config |
30
|
|
|
*/ |
31
|
|
|
private static $server_endpoint; |
|
|
|
|
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* @var TikaRestClient |
35
|
|
|
*/ |
36
|
|
|
protected $client = null; |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* Cache of supported mime types |
40
|
|
|
* |
41
|
|
|
* @var array |
42
|
|
|
*/ |
43
|
|
|
protected $supportedMimes = []; |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* @return TikaRestClient |
47
|
|
|
*/ |
48
|
|
|
public function getClient() |
49
|
|
|
{ |
50
|
|
|
if (!$this->client) { |
51
|
|
|
$this->client = Injector::inst()->createWithArgs( |
52
|
|
|
TikaRestClient::class, |
53
|
|
|
[$this->getServerEndpoint()] |
54
|
|
|
); |
55
|
|
|
} |
56
|
|
|
return $this->client; |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* @return string |
61
|
|
|
*/ |
62
|
|
|
public function getServerEndpoint() |
63
|
|
|
{ |
64
|
|
|
if ($endpoint = Environment::getEnv('SS_TIKA_ENDPOINT')) { |
65
|
|
|
return $endpoint; |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
// Default to configured endpoint |
69
|
|
|
return $this->config()->get('server_endpoint'); |
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* Get the version of Tika installed, or 0 if not installed |
74
|
|
|
* |
75
|
|
|
* @return float version of Tika |
76
|
|
|
*/ |
77
|
|
|
public function getVersion() |
78
|
|
|
{ |
79
|
|
|
return $this->getClient()->getVersion(); |
|
|
|
|
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* @return boolean |
84
|
|
|
*/ |
85
|
|
|
public function isAvailable() |
86
|
|
|
{ |
87
|
|
|
return $this->getServerEndpoint() |
88
|
|
|
&& $this->getClient()->isAvailable() |
89
|
|
|
&& version_compare($this->getVersion(), '1.7') >= 0; |
90
|
|
|
} |
91
|
|
|
|
92
|
|
|
/** |
93
|
|
|
* @param string $extension |
94
|
|
|
* @return boolean |
95
|
|
|
*/ |
96
|
|
|
public function supportsExtension($extension) |
97
|
|
|
{ |
98
|
|
|
// Determine support via mime type only |
99
|
|
|
return false; |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
/** |
103
|
|
|
* @param string $mime |
104
|
|
|
* @return boolean |
105
|
|
|
*/ |
106
|
|
|
public function supportsMime($mime) |
107
|
|
|
{ |
108
|
|
|
if (!$this->supportedMimes) { |
|
|
|
|
109
|
|
|
$this->supportedMimes = (array) $this->getClient()->getSupportedMimes(); |
110
|
|
|
} |
111
|
|
|
|
112
|
|
|
// Check if supported (most common / quickest lookup) |
113
|
|
|
if (isset($this->supportedMimes[$mime])) { |
114
|
|
|
return true; |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
// Check aliases |
118
|
|
|
foreach ($this->supportedMimes as $info) { |
119
|
|
|
if (isset($info['alias']) && in_array($mime, $info['alias'])) { |
120
|
|
|
return true; |
121
|
|
|
} |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
return false; |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
public function getContent($file) |
128
|
|
|
{ |
129
|
|
|
$tempFile = $file instanceof File ? $this->getPathFromFile($file) : $file; |
130
|
|
|
$content = $this->getClient()->tika($tempFile); |
131
|
|
|
//Cleanup temp file |
132
|
|
|
if ($file instanceof File) { |
133
|
|
|
unlink($tempFile); |
134
|
|
|
} |
135
|
|
|
return $content; |
136
|
|
|
} |
137
|
|
|
} |
138
|
|
|
|