|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Matecat\SubFiltering\Filters; |
|
4
|
|
|
|
|
5
|
|
|
use Exception; |
|
6
|
|
|
use Matecat\SubFiltering\Commons\AbstractHandler; |
|
7
|
|
|
use Matecat\SubFiltering\Enum\CTypeEnum; |
|
8
|
|
|
use Matecat\SubFiltering\Utils\DataRefReplacer; |
|
9
|
|
|
|
|
10
|
|
|
class DataRefRestore extends AbstractHandler { |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* @var array |
|
14
|
|
|
*/ |
|
15
|
|
|
private array $dataRefMap = []; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* DataRefReplace constructor. |
|
19
|
|
|
*/ |
|
20
|
38 |
|
public function __construct() { |
|
21
|
38 |
|
parent::__construct(); |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @inheritDoc |
|
26
|
|
|
* @throws Exception |
|
27
|
|
|
*/ |
|
28
|
38 |
|
public function transform( string $segment ): string { |
|
29
|
|
|
|
|
30
|
38 |
|
if ( empty( $this->dataRefMap ) ) { |
|
31
|
38 |
|
$this->dataRefMap = $this->pipeline->getDataRefMap(); |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
38 |
|
if ( empty( $this->dataRefMap ) ) { |
|
35
|
33 |
|
$segment = $this->restoreXliffPhTagsFromMatecatPhTags( $segment ); |
|
36
|
|
|
|
|
37
|
33 |
|
return $this->restoreXliffPcTagsFromMatecatPhTags( $segment ); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
5 |
|
$dataRefReplacer = new DataRefReplacer( $this->dataRefMap ); |
|
41
|
5 |
|
$segment = $dataRefReplacer->restore( $segment ); |
|
42
|
5 |
|
$segment = $this->restoreXliffPhTagsFromMatecatPhTags( $segment ); |
|
43
|
|
|
|
|
44
|
5 |
|
return $this->restoreXliffPcTagsFromMatecatPhTags( $segment ); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* This function restores <ph> tags (encoded as Matecat ph tags) without any dataRef correspondence |
|
49
|
|
|
* for the persistence layer (layer 0). |
|
50
|
|
|
* |
|
51
|
|
|
* Example: |
|
52
|
|
|
* |
|
53
|
|
|
* Saame nähtavuse piirangutega kontrollida, kes sisu näeb .<ph id="mtc_ph_u_1" equiv-text="base64:Jmx0O3BoIGlkPSJzb3VyY2UxIiBkYXRhUmVmPSJzb3VyY2UxIi8mZ3Q7"/> |
|
54
|
|
|
* |
|
55
|
|
|
* is transformed to: |
|
56
|
|
|
* |
|
57
|
|
|
* Saame nähtavuse piirangutega kontrollida, kes sisu näeb .<ph id="source1" dataRef="source1"/> |
|
58
|
|
|
* |
|
59
|
|
|
* @param string $segment |
|
60
|
|
|
* |
|
61
|
|
|
* @return string |
|
62
|
|
|
*/ |
|
63
|
38 |
|
private function restoreXliffPhTagsFromMatecatPhTags( string $segment ): string { |
|
64
|
38 |
|
preg_match_all( '|<ph[^>]+? ctype="' . CTypeEnum::ORIGINAL_PH_OR_NOT_DATA_REF . '" equiv-text="base64:(.*?)"/>|iu', $segment, $matches ); |
|
65
|
|
|
|
|
66
|
38 |
|
if ( empty( $matches[ 0 ] ) ) { |
|
67
|
34 |
|
return $segment; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
4 |
|
foreach ( $matches[ 0 ] as $index => $match ) { |
|
71
|
4 |
|
$segment = str_replace( $match, base64_decode( $matches[ 1 ][ $index ] ), $segment ); |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
4 |
|
return $segment; |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
/** |
|
78
|
|
|
* This function restores <pc> tags (encoded as Matecat ph tags) without any dataRef correspondence |
|
79
|
|
|
* for the persistence layer (layer 0). |
|
80
|
|
|
* |
|
81
|
|
|
* Example: |
|
82
|
|
|
* |
|
83
|
|
|
* Testo <ph id="source1_1" dataType="pcStart" originalData="Jmx0O3BjIGlkPSJzb3VyY2UxIiBkYXRhUmVmU3RhcnQ9InNvdXJjZTEiIGRhdGFSZWZFbmQ9InNvdXJjZTEiJmd0Ow==" dataRef="source1" equiv-text="base64:Xw=="/><ph id="mtc_u_1" equiv-text="base64:Jmx0O3BjIGlkPSIxdSIgdHlwZT0iZm10IiBzdWJUeXBlPSJtOnUiJmd0Ow=="/><ph id="mtc_u_2" equiv-text="base64:Jmx0Oy9wYyZndDs="/><ph id="source1_2" dataType="pcEnd" originalData="Jmx0Oy9wYyZndDs=" dataRef="source1" equiv-text="base64:Xw=="/> |
|
84
|
|
|
* |
|
85
|
|
|
* is transformed to: |
|
86
|
|
|
* |
|
87
|
|
|
* Testo <pc id="source1" dataRefStart="source1" dataRefEnd="source1"><pc id="1u" type="fmt" subType="m:u"></pc></pc> |
|
88
|
|
|
* |
|
89
|
|
|
* @param $segment |
|
90
|
|
|
* |
|
91
|
|
|
* @return string |
|
92
|
|
|
*/ |
|
93
|
38 |
|
private function restoreXliffPcTagsFromMatecatPhTags( string $segment ): string { |
|
94
|
|
|
|
|
95
|
38 |
|
$matches = []; |
|
96
|
38 |
|
preg_match_all( '|<ph[^>]+? ctype="' . CTypeEnum::ORIGINAL_PC_OPEN_NO_DATA_REF . '" equiv-text="base64:(.*?)"/>|iu', $segment, $open_matches ); |
|
97
|
38 |
|
preg_match_all( '|<ph[^>]+? ctype="' . CTypeEnum::ORIGINAL_PC_CLOSE_NO_DATA_REF . '" equiv-text="base64:(.*?)"/>|iu', $segment, $close_matches ); |
|
98
|
|
|
|
|
99
|
38 |
|
$matches[ 0 ] = array_merge( $open_matches[ 0 ], $close_matches[ 0 ] ); |
|
100
|
38 |
|
$matches[ 1 ] = array_merge( $open_matches[ 1 ], $close_matches[ 1 ] ); |
|
101
|
|
|
|
|
102
|
38 |
|
if ( empty( $matches[ 0 ] ) ) { |
|
103
|
36 |
|
return $segment; |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
2 |
|
foreach ( $matches[ 0 ] as $index => $match ) { |
|
107
|
2 |
|
$segment = str_replace( $match, base64_decode( $matches[ 1 ][ $index ] ), $segment ); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
2 |
|
return $segment; |
|
111
|
|
|
} |
|
112
|
|
|
} |