|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Vaites\ApacheTika\Metadata; |
|
4
|
|
|
|
|
5
|
|
|
use DateTime; |
|
6
|
|
|
use DateTimeZone; |
|
7
|
|
|
use Exception; |
|
8
|
|
|
use stdClass; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Standarized metadata class with common attributes for all document types |
|
12
|
|
|
* |
|
13
|
|
|
* @author David Martínez <[email protected]> |
|
14
|
|
|
*/ |
|
15
|
|
|
abstract class Metadata implements MetadataInterface |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* Title |
|
19
|
|
|
* |
|
20
|
|
|
* @var string |
|
21
|
|
|
*/ |
|
22
|
|
|
public $title = null; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* Content |
|
26
|
|
|
* |
|
27
|
|
|
* @var string |
|
28
|
|
|
*/ |
|
29
|
|
|
public $content = null; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* MIME type |
|
33
|
|
|
* |
|
34
|
|
|
* @var string |
|
35
|
|
|
*/ |
|
36
|
|
|
public $mime = null; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Date created |
|
40
|
|
|
* |
|
41
|
|
|
* @var \DateTime |
|
42
|
|
|
*/ |
|
43
|
|
|
public $created = null; |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* Date updated or last modified |
|
47
|
|
|
* |
|
48
|
|
|
* @var \DateTime |
|
49
|
|
|
*/ |
|
50
|
|
|
public $updated = null; |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* RAW attributes returned by Apache Tika |
|
54
|
|
|
* |
|
55
|
|
|
* @var \stdClass |
|
56
|
|
|
*/ |
|
57
|
|
|
public $meta = null; |
|
58
|
|
|
|
|
59
|
|
|
/** |
|
60
|
|
|
* Parse Apache Tika response filling all properties |
|
61
|
|
|
* |
|
62
|
|
|
* @throws \Exception |
|
63
|
|
|
*/ |
|
64
|
|
|
public function __construct(stdClass $meta, string $file) |
|
65
|
|
|
{ |
|
66
|
|
|
$this->meta = $meta; |
|
67
|
|
|
|
|
68
|
|
|
// process each meta |
|
69
|
|
|
foreach((array) $this->meta as $key => $value) |
|
70
|
|
|
{ |
|
71
|
|
|
$this->setAttribute($key, $value); |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
// file name without extension if title is not detected |
|
75
|
|
|
if(empty($this->title)) |
|
76
|
|
|
{ |
|
77
|
|
|
$this->title = (string) preg_replace('/\..+$/', '', basename($file)); |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
// use creation date as last modified if not detected |
|
81
|
|
|
if(empty($this->updated)) |
|
82
|
|
|
{ |
|
83
|
|
|
$this->updated = $this->created; |
|
84
|
|
|
} |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
/** |
|
88
|
|
|
* Return an instance of Metadata based on content type |
|
89
|
|
|
* |
|
90
|
|
|
* @throws \Exception |
|
91
|
|
|
*/ |
|
92
|
|
|
public static function make(stdClass $meta, string $file): MetadataInterface |
|
93
|
|
|
{ |
|
94
|
|
|
// get content type |
|
95
|
|
|
$mime = is_array($meta->{'Content-Type'}) ? current($meta->{'Content-Type'}) : $meta->{'Content-Type'}; |
|
96
|
|
|
|
|
97
|
|
|
// instance based on content type |
|
98
|
|
|
switch(current(explode('/', $mime))) |
|
99
|
|
|
{ |
|
100
|
|
|
case 'image': |
|
101
|
|
|
$instance = new ImageMetadata($meta, $file); |
|
102
|
|
|
break; |
|
103
|
|
|
|
|
104
|
|
|
default: |
|
105
|
|
|
$instance = new DocumentMetadata($meta, $file); |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
return $instance; |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* Sets an attribute |
|
113
|
|
|
* |
|
114
|
|
|
* @param mixed $value |
|
115
|
|
|
* @return \Vaites\ApacheTika\Metadata\MetadataInterface |
|
116
|
|
|
* @throws \Exception |
|
117
|
|
|
*/ |
|
118
|
|
|
public final function setAttribute(string $key, $value): MetadataInterface |
|
119
|
|
|
{ |
|
120
|
|
|
$timezone = new DateTimeZone('UTC'); |
|
121
|
|
|
|
|
122
|
|
|
switch(mb_strtolower($key)) |
|
123
|
|
|
{ |
|
124
|
|
|
case 'content-type': |
|
125
|
|
|
$mime = $value ? (preg_split('/;\s+/', $value) ?: []) : []; |
|
126
|
|
|
|
|
127
|
|
|
if(count($mime)) |
|
128
|
|
|
{ |
|
129
|
|
|
$this->mime = array_shift($mime); |
|
130
|
|
|
} |
|
131
|
|
|
break; |
|
132
|
|
|
|
|
133
|
|
|
case 'creation-date': |
|
134
|
|
|
case 'date': |
|
135
|
|
|
case 'dcterms:created': |
|
136
|
|
|
case 'meta:creation-date': |
|
137
|
|
|
$value = preg_replace('/\.\d+/', 'Z', $value); |
|
138
|
|
|
$this->created = new DateTime(is_array($value) ? array_shift($value) : $value, $timezone); |
|
139
|
|
|
break; |
|
140
|
|
|
|
|
141
|
|
|
case 'dcterms:modified': |
|
142
|
|
|
case 'last-modified': |
|
143
|
|
|
case 'modified': |
|
144
|
|
|
$value = preg_replace('/\.\d+/', 'Z', $value); |
|
145
|
|
|
$this->updated = new DateTime(is_array($value) ? array_shift($value) : $value, $timezone); |
|
146
|
|
|
break; |
|
147
|
|
|
|
|
148
|
|
|
default: |
|
149
|
|
|
$this->setSpecificAttribute($key, $value); |
|
150
|
|
|
|
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
return $this; |
|
154
|
|
|
} |
|
155
|
|
|
|
|
156
|
|
|
/** |
|
157
|
|
|
* Sets an speficic attribute for the file type |
|
158
|
|
|
* |
|
159
|
|
|
* @param mixed $value |
|
160
|
|
|
* @return \Vaites\ApacheTika\Metadata\MetadataInterface |
|
161
|
|
|
*/ |
|
162
|
|
|
abstract protected function setSpecificAttribute(string $key, $value): MetadataInterface; |
|
163
|
|
|
} |
|
164
|
|
|
|
vaites /
php-apache-tika
