|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* (c) Kitodo. Key to digital objects e.V. <[email protected]> |
|
5
|
|
|
* |
|
6
|
|
|
* This file is part of the Kitodo and TYPO3 projects. |
|
7
|
|
|
* |
|
8
|
|
|
* @license GNU General Public License version 3 or later. |
|
9
|
|
|
* For the full copyright and license information, please read the |
|
10
|
|
|
* LICENSE.txt file that was distributed with this source code. |
|
11
|
|
|
*/ |
|
12
|
|
|
|
|
13
|
|
|
namespace Kitodo\Dlf\Tests\Unit\Format; |
|
14
|
|
|
|
|
15
|
|
|
use Kitodo\Dlf\Format\Alto; |
|
16
|
|
|
use TYPO3\TestingFramework\Core\Unit\UnitTestCase; |
|
17
|
|
|
|
|
18
|
|
|
class AltoTest extends UnitTestCase |
|
19
|
|
|
{ |
|
20
|
|
|
/** |
|
21
|
|
|
* @test |
|
22
|
|
|
* @group extract data |
|
23
|
|
|
*/ |
|
24
|
|
|
public function getRawData(): void |
|
25
|
|
|
{ |
|
26
|
|
|
$xml = simplexml_load_file(__DIR__ . '/../../Fixtures/Format/alto.xml'); |
|
27
|
|
|
$alto = new Alto(); |
|
28
|
|
|
|
|
29
|
|
|
$rawText = $alto->getRawText($xml); |
|
30
|
|
|
|
|
31
|
|
|
self::assertEquals('Bürgertum und Bürgerlichkeit in Dresden DRESDNER HEFTE', $rawText); |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* @test |
|
36
|
|
|
* @group extract data |
|
37
|
|
|
*/ |
|
38
|
|
|
public function getTextAsMiniOcr(): void |
|
39
|
|
|
{ |
|
40
|
|
|
$xml = simplexml_load_file(__DIR__ . '/../../Fixtures/Format/alto.xml'); |
|
41
|
|
|
$alto = new Alto(); |
|
42
|
|
|
|
|
43
|
|
|
$rawText = $alto->getTextAsMiniOcr($xml); |
|
44
|
|
|
|
|
45
|
|
|
$miniOCR = <<<XML |
|
46
|
|
|
<ocr> |
|
47
|
|
|
<b> |
|
48
|
|
|
<l> |
|
49
|
|
|
<w x="477 2083 437 95">Bürgertum </w> |
|
50
|
|
|
<w x="950 2076 155 76">und </w> |
|
51
|
|
|
</l> |
|
52
|
|
|
<l> |
|
53
|
|
|
<w x="477 2201 574 102">Bürgerlichkeit </w> |
|
54
|
|
|
<w x="1084 2205 74 68">in </w> |
|
55
|
|
|
<w x="1194 2199 333 75">Dresden </w> |
|
56
|
|
|
</l> |
|
57
|
|
|
</b> |
|
58
|
|
|
<b> |
|
59
|
|
|
<l> |
|
60
|
|
|
<w x="473 315 752 98">DRESDNER </w> |
|
61
|
|
|
</l> |
|
62
|
|
|
<l> |
|
63
|
|
|
<w x="473 492 448 97">HEFTE </w> |
|
64
|
|
|
</l> |
|
65
|
|
|
</b> |
|
66
|
|
|
</ocr> |
|
67
|
|
|
XML; |
|
68
|
|
|
|
|
69
|
|
|
self::assertXmlStringEqualsXmlString($miniOCR, $rawText); |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* @test |
|
74
|
|
|
* @group extract data |
|
75
|
|
|
*/ |
|
76
|
|
|
public function getTextAsMiniOcrNoTextBlock(): void |
|
77
|
|
|
{ |
|
78
|
|
|
$xml = simplexml_load_file(__DIR__ . '/../../Fixtures/Format/altoNoTextBlock.xml'); |
|
79
|
|
|
$alto = new Alto(); |
|
80
|
|
|
|
|
81
|
|
|
$rawText = $alto->getTextAsMiniOcr($xml); |
|
82
|
|
|
|
|
83
|
|
|
self::assertEquals('', $rawText); |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
/** |
|
87
|
|
|
* @test |
|
88
|
|
|
* @group extract data |
|
89
|
|
|
*/ |
|
90
|
|
|
public function getTextAsMiniOcrNoTextline(): void |
|
91
|
|
|
{ |
|
92
|
|
|
$xml = simplexml_load_file(__DIR__ . '/../../Fixtures/Format/altoNoTextLine.xml'); |
|
93
|
|
|
$alto = new Alto(); |
|
94
|
|
|
|
|
95
|
|
|
$rawText = $alto->getTextAsMiniOcr($xml); |
|
96
|
|
|
|
|
97
|
|
|
self::assertXmlStringEqualsXmlString('<?xml version="1.0"?><ocr><b/><b/></ocr>', $rawText); |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* @test |
|
102
|
|
|
* @group extract data |
|
103
|
|
|
*/ |
|
104
|
|
|
public function getTextAsMiniOcrNoString(): void |
|
105
|
|
|
{ |
|
106
|
|
|
$xml = simplexml_load_file(__DIR__ . '/../../Fixtures/Format/altoNoString.xml'); |
|
107
|
|
|
$alto = new Alto(); |
|
108
|
|
|
|
|
109
|
|
|
$rawText = $alto->getTextAsMiniOcr($xml); |
|
110
|
|
|
|
|
111
|
|
|
self::assertXmlStringEqualsXmlString( |
|
112
|
|
|
'<?xml version="1.0"?><ocr><b><l/><l/></b><b><l/><l/></b></ocr>', |
|
113
|
|
|
$rawText |
|
114
|
|
|
); |
|
115
|
|
|
} |
|
116
|
|
|
} |
|
117
|
|
|
|