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