author | ymh <ymh.work@gmail.com> |
Wed, 15 Feb 2017 10:51:15 +0100 | |
changeset 509 | fcc59d0ac8aa |
parent 163 | 59c68fc4848e |
permissions | -rw-r--r-- |
162
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
1 |
<?php |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
2 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
3 |
use Mockery as m; |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
4 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
5 |
use CorpusParole\Models\Document; |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
6 |
use CorpusParole\Libraries\Transcript\LacitoTranscriptConverter; |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
7 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
8 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
9 |
/** |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
10 |
* |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
11 |
*/ |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
12 |
class LacitoTranscriptConverterTest extends TestCase { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
13 |
const TEST_DOC_BASE = "crdo-UVE_MOCIKA"; |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
14 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
15 |
public function setUp() { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
16 |
parent::setup(); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
17 |
$graphContent = sprintf(file_get_contents(__DIR__ . DIRECTORY_SEPARATOR . self::TEST_DOC_BASE.".ttl"), config('corpusparole.corpus_doc_id_base_uri'), config('corpusparole.corpus_id_scheme')); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
18 |
$this->graph = new EasyRdf\Graph(config('corpusparole.corpus_doc_id_base_uri')."crdo-UVE_MOCIKA_SOUND", $graphContent); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
19 |
$this->doc = new Document(config('corpusparole.corpus_doc_id_base_uri')."crdo-UVE_MOCIKA_SOUND", $this->graph); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
20 |
|
163 | 21 |
$this->transcriptSource = file_get_contents(__DIR__ . DIRECTORY_SEPARATOR . self::TEST_DOC_BASE.".xml"); |
162
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
22 |
|
163 | 23 |
$this->transcriptSourceSpeaker = file_get_contents(__DIR__ . DIRECTORY_SEPARATOR ."crdo-FRA_PK_IV_10.xml"); |
162
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
24 |
|
163 | 25 |
$this->transcriptSourceNoContent = file_get_contents(__DIR__ . DIRECTORY_SEPARATOR ."crdo-FSL-CUC023.xml"); |
162
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
26 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
27 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
28 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
29 |
public function getMockConverter(...$contructorArgs) { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
30 |
return new LacitoTranscriptConverter(...$contructorArgs); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
31 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
32 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
33 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
34 |
public function testConstructor() { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
35 |
$converter = $this->getMockConverter($this->doc, $this->transcriptSource); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
36 |
$json = $converter->convertToJson(); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
37 |
$this->assertNotnull($json); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
38 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
39 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
40 |
public function testGetSourceTitle() { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
41 |
$converter = $this->getMockConverter($this->doc, $this->transcriptSource); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
42 |
$titles = $converter->getSourceTitle(); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
43 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
44 |
$this->assertEquals(['en' => 'The two hermit crabs and the coconut crab', 'fr' => "Les deux bernard-l'hermite et le crabe de cocotier"], $titles, "Titles must come from the xml file"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
45 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
46 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
47 |
public function testBuildAnnotations() { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
48 |
$converter = $this->getMockConverter($this->doc, $this->transcriptSource); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
49 |
$annotations = $converter->buildAnnotations(); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
50 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
51 |
$this->assertCount(32, $annotations, "Must have 32 annotation"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
52 |
foreach($annotations as $i => $a) { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
53 |
$this->assertArrayHasKey('begin', $a, "Must have begin key"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
54 |
$this->assertArrayHasKey('end', $a, "Must have begin key"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
55 |
$this->assertTrue($a['begin']<$a['end'], "Begin is < to endP"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
56 |
$this->assertEquals($this->doc->getId()."_a".sprintf("%03s",($i+1)), $a['id']); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
57 |
$this->assertEquals($converter->getMediaRefId(), $a['media']); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
58 |
$this->assertArrayHasKey('content', $a, "must have content"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
59 |
$this->assertTrue(is_array($a['content'])); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
60 |
$this->assertArrayHasKey('mimetype', $a['content']); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
61 |
$this->assertEquals('application/json', $a['content']['mimetype']); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
62 |
$this->assertArrayHasKey('data', $a['content']); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
63 |
$this->assertTrue(is_array($a['content']['data'])); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
64 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
65 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
66 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
67 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
68 |
public function testBuildAnnotationsContent() { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
69 |
$converter = $this->getMockConverter($this->doc, $this->transcriptSource); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
70 |
$annotations = $converter->buildAnnotations(); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
71 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
72 |
foreach($annotations as $i => $a) { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
73 |
$data = $a['content']['data']; |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
74 |
$this->assertArrayNotHasKey('speaker', $data, "No spreaker in data for this source"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
75 |
$this->assertArrayHasKey('content', $data, "data has content"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
76 |
$this->assertNotEmpty($data['content'], "content not empty"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
77 |
$this->arrayHasKey('transl', $data, 'data has translation'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
78 |
$this->assertTrue(is_array($data['transl']), 'data transl is array'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
79 |
$this->assertArrayHasKey('@value', $data['transl'], 'dats transl has @value key'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
80 |
$this->assertArrayHasKey('@language', $data['transl'], 'dats transl has @language key'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
81 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
82 |
$this->assertArrayHasKey('words', $data, 'data has words'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
83 |
$this->assertNotEmpty($data['words'], "words are not empty"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
84 |
foreach($data['words'] as $w) { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
85 |
$this->assertTrue(is_array($w), 'words are array'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
86 |
$this->assertArrayHasKey('content', $w, "words have content"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
87 |
$this->assertNotEmpty($w['content'], "words have non empty content"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
88 |
$this->assertArrayHasKey('transl', $w, "words have transl"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
89 |
$this->assertArrayHasKey('@value', $w['transl'], 'words transl has @value key'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
90 |
$this->assertArrayHasKey('@language', $w['transl'], 'words transl has @language key'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
91 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
92 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
93 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
94 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
95 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
96 |
public function testBuildAnnotationsSpeaker() { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
97 |
$converter = $this->getMockConverter($this->doc, $this->transcriptSourceSpeaker); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
98 |
$annotations = $converter->buildAnnotations(); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
99 |
foreach($annotations as $i => $a) { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
100 |
$data = $a['content']['data']; |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
101 |
$this->assertArrayHasKey('speaker', $data, 'annotation must have speaker'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
102 |
$this->assertArrayNotHasKey('words', $data, 'No words in data here'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
103 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
104 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
105 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
106 |
public function testBuildAnnotationsSpeakerTitle() { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
107 |
$converter = $this->getMockConverter($this->doc, $this->transcriptSourceSpeaker); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
108 |
$this->assertNull($converter->getSourceTitle(), "Title is not on source"); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
109 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
110 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
111 |
public function testBuildAnnotationsNoContent() { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
112 |
$converter = $this->getMockConverter($this->doc, $this->transcriptSourceNoContent); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
113 |
$annotations = $converter->buildAnnotations(); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
114 |
foreach($annotations as $i => $a) { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
115 |
$data = $a['content']['data']; |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
116 |
$this->assertArrayNotHasKey('words', $data, 'No words in data here'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
117 |
$this->assertArrayNotHasKey('content', $data, 'No content in data here'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
118 |
$this->arrayHasKey('transl', $data, 'data has translation'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
119 |
$this->assertTrue(is_array($data['transl']), 'data transl is array'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
120 |
$this->assertArrayHasKey('@value', $data['transl'], 'dats transl has @value key'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
121 |
$this->assertArrayHasKey('@language', $data['transl'], 'dats transl has @language key'); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
122 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
123 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
124 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
125 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
126 |
public function tearDown() { |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
127 |
m::close(); |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
128 |
} |
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
129 |
|
a6cf5a06f02d
add Transcript converters libraries + test
ymh <ymh.work@gmail.com>
parents:
diff
changeset
|
130 |
} |