|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Matecat\XliffParser\XliffParser; |
|
4
|
|
|
|
|
5
|
|
|
use DOMAttr; |
|
6
|
|
|
use DOMDocument; |
|
7
|
|
|
use DOMElement; |
|
8
|
|
|
use DOMNode; |
|
9
|
|
|
use Exception; |
|
10
|
|
|
use Matecat\XliffParser\Exception\DuplicateTransUnitIdInXliff; |
|
11
|
|
|
use Matecat\XliffParser\Exception\NotFoundIdInTransUnit; |
|
12
|
|
|
use Matecat\XliffParser\Exception\SegmentIdTooLongException; |
|
13
|
|
|
|
|
14
|
|
|
class XliffParserV1 extends AbstractXliffParser { |
|
15
|
|
|
/** |
|
16
|
|
|
* @inheritDoc |
|
17
|
|
|
* @throws Exception |
|
18
|
|
|
*/ |
|
19
|
|
|
public function parse( DOMDocument $dom, $output = [] ) { |
|
20
|
|
|
$i = 1; |
|
21
|
|
|
/** @var DOMElement $file */ |
|
22
|
|
|
foreach ( $dom->getElementsByTagName( 'file' ) as $file ) { |
|
23
|
|
|
|
|
24
|
|
|
// metadata |
|
25
|
|
|
$output[ 'files' ][ $i ][ 'attr' ] = $this->extractMetadata( $file ); |
|
26
|
|
|
|
|
27
|
|
|
// reference |
|
28
|
|
|
if ( !empty( $this->extractReference( $file ) ) ) { |
|
29
|
|
|
$output[ 'files' ][ $i ][ 'reference' ] = $this->extractReference( $file ); |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
// trans-units |
|
33
|
|
|
$transUnitIdArrayForUniquenessCheck = []; |
|
34
|
|
|
$j = 1; |
|
35
|
|
|
foreach ( $file->childNodes as $body ) { |
|
36
|
|
|
if ( $body->nodeName === 'body' ) { |
|
37
|
|
|
foreach ( $body->childNodes as $childNode ) { |
|
38
|
|
|
$this->extractTuFromNode( $childNode, $transUnitIdArrayForUniquenessCheck, $dom, $output, $i, $j ); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
// trans-unit re-count check |
|
42
|
|
|
$totalTransUnitsId = count( $transUnitIdArrayForUniquenessCheck ); |
|
43
|
|
|
$transUnitsUniqueId = count( array_unique( $transUnitIdArrayForUniquenessCheck ) ); |
|
44
|
|
|
if ( $totalTransUnitsId != $transUnitsUniqueId ) { |
|
45
|
|
|
throw new DuplicateTransUnitIdInXliff( "Invalid trans-unit id, duplicate found.", 400 ); |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
$i++; |
|
49
|
|
|
} |
|
50
|
|
|
} |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
return $output; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* @param DOMElement $file |
|
58
|
|
|
* |
|
59
|
|
|
* @return array |
|
60
|
|
|
*/ |
|
61
|
|
|
private function extractMetadata( DOMElement $file ) { |
|
62
|
|
|
$metadata = []; |
|
63
|
|
|
$customAttr = []; |
|
64
|
|
|
|
|
65
|
|
|
/** @var DOMAttr $attribute */ |
|
66
|
|
|
foreach ( $file->attributes as $attribute ) { |
|
67
|
|
|
switch ( $attribute->localName ) { |
|
68
|
|
|
// original |
|
69
|
|
|
case 'original': |
|
70
|
|
|
$metadata[ 'original' ] = $attribute->value; |
|
71
|
|
|
break; |
|
72
|
|
|
|
|
73
|
|
|
// source-language |
|
74
|
|
|
case 'source-language': |
|
75
|
|
|
$metadata[ 'source-language' ] = $attribute->value; |
|
76
|
|
|
break; |
|
77
|
|
|
|
|
78
|
|
|
// data-type |
|
79
|
|
|
case 'datatype': |
|
80
|
|
|
$metadata[ 'data-type' ] = $attribute->value; |
|
81
|
|
|
break; |
|
82
|
|
|
|
|
83
|
|
|
// target-language |
|
84
|
|
|
case 'target-language': |
|
85
|
|
|
$metadata[ 'target-language' ] = $attribute->value; |
|
86
|
|
|
break; |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
// Custom MateCat x-Attribute |
|
90
|
|
|
preg_match( '|x-(.*?)|si', $attribute->localName, $temp ); |
|
91
|
|
|
if ( isset( $temp[ 1 ] ) ) { |
|
92
|
|
|
$customAttr[ $attribute->localName ] = $attribute->value; |
|
93
|
|
|
} |
|
94
|
|
|
unset( $temp ); |
|
95
|
|
|
|
|
96
|
|
|
// Custom MateCat namespace Attribute mtc: |
|
97
|
|
|
preg_match( '|mtc:(.*?)|si', $attribute->nodeName, $temp ); |
|
98
|
|
|
if ( isset( $temp[ 1 ] ) ) { |
|
99
|
|
|
$customAttr[ $attribute->nodeName ] = $attribute->value; |
|
100
|
|
|
} |
|
101
|
|
|
unset( $temp ); |
|
102
|
|
|
|
|
103
|
|
|
if ( !empty( $customAttr ) ) { |
|
104
|
|
|
$metadata[ 'custom' ] = $customAttr; |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
return $metadata; |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* @param DOMElement $file |
|
113
|
|
|
* |
|
114
|
|
|
* @return array |
|
115
|
|
|
*/ |
|
116
|
|
|
private function extractReference( DOMElement $file ) { |
|
117
|
|
|
$reference = []; |
|
118
|
|
|
|
|
119
|
|
|
$order = 0; |
|
120
|
|
|
foreach ( $file->getElementsByTagName( 'reference' ) as $ref ) { |
|
121
|
|
|
/** @var DOMNode $childNode */ |
|
122
|
|
|
foreach ( $ref->childNodes as $childNode ) { |
|
123
|
|
|
if ( $childNode->nodeName === 'internal-file' ) { |
|
124
|
|
|
$reference[ $order ][ 'form-type' ] = $childNode->attributes->getNamedItem( 'form' )->nodeValue; |
|
125
|
|
|
$reference[ $order ][ 'base64' ] = trim( $childNode->nodeValue ); |
|
126
|
|
|
$order++; |
|
127
|
|
|
} |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
return $reference; |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
/** |
|
135
|
|
|
* Extract and populate 'trans-units' array |
|
136
|
|
|
* |
|
137
|
|
|
* @param $transUnit |
|
138
|
|
|
* @param $transUnitIdArrayForUniquenessCheck |
|
139
|
|
|
* @param $dom |
|
140
|
|
|
* @param $output |
|
141
|
|
|
* @param $i |
|
142
|
|
|
* @param $j |
|
143
|
|
|
* |
|
144
|
|
|
* @throws Exception |
|
145
|
|
|
*/ |
|
146
|
|
|
protected function extractTransUnit( $transUnit, &$transUnitIdArrayForUniquenessCheck, $dom, &$output, &$i, &$j ) { |
|
147
|
|
|
// metadata |
|
148
|
|
|
$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'attr' ] = $this->extractTransUnitMetadata( $transUnit, $transUnitIdArrayForUniquenessCheck ); |
|
149
|
|
|
|
|
150
|
|
|
// notes |
|
151
|
|
|
$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'notes' ] = $this->extractTransUnitNotes( $dom, $transUnit ); |
|
152
|
|
|
|
|
153
|
|
|
// content |
|
154
|
|
|
/** @var DOMElement $childNode */ |
|
155
|
|
|
foreach ( $transUnit->childNodes as $childNode ) { |
|
156
|
|
|
// source |
|
157
|
|
|
if ( $childNode->nodeName === 'source' ) { |
|
158
|
|
|
$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'source' ] = $this->extractContent( $dom, $childNode ); |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
// seg-source |
|
162
|
|
|
if ( $childNode->nodeName === 'seg-source' ) { |
|
163
|
|
|
$rawSegment = $output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'source' ][ 'raw-content' ]; |
|
164
|
|
|
$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'seg-source' ] = $this->extractContentWithMarksAndExtTags( $dom, $childNode, $rawSegment ); |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
// target |
|
168
|
|
|
if ( $childNode->nodeName === 'target' ) { |
|
169
|
|
|
$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'target' ] = $this->extractContent( $dom, $childNode ); |
|
170
|
|
|
|
|
171
|
|
|
// seg-target |
|
172
|
|
|
$targetRawContent = @$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'target' ][ 'raw-content' ]; |
|
173
|
|
|
$segSource = @$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'seg-source' ]; |
|
174
|
|
|
if ( isset( $targetRawContent ) && !empty( $targetRawContent ) && isset( $segSource ) && count( $segSource ) > 0 ) { |
|
175
|
|
|
$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'seg-target' ] = $this->extractContentWithMarksAndExtTags( $dom, $childNode, $targetRawContent ); |
|
176
|
|
|
} |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
// locked |
|
180
|
|
|
if ( $childNode->nodeName === 'sdl:seg' ) { |
|
181
|
|
|
$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'locked' ] = $this->extractLocked( $childNode ); |
|
182
|
|
|
} |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
|
|
// context-group |
|
186
|
|
|
foreach ( $transUnit->getElementsByTagName( 'context-group' ) as $contextGroup ) { |
|
187
|
|
|
$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'context-group' ][] = $this->extractTransUnitContextGroup( $dom, $contextGroup ); |
|
188
|
|
|
} |
|
189
|
|
|
|
|
190
|
|
|
// alt-trans |
|
191
|
|
|
foreach ( $transUnit->getElementsByTagName( 'alt-trans' ) as $altTrans ) { |
|
192
|
|
|
$output[ 'files' ][ $i ][ 'trans-units' ][ $j ][ 'alt-trans' ][] = $this->extractTransUnitAltTrans( $altTrans ); |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
$j++; |
|
196
|
|
|
} |
|
197
|
|
|
|
|
198
|
|
|
/** |
|
199
|
|
|
* @param DOMElement $transUnit |
|
200
|
|
|
* @param array $transUnitIdArrayForUniquenessCheck |
|
201
|
|
|
* |
|
202
|
|
|
* @return array |
|
203
|
|
|
* @throws Exception |
|
204
|
|
|
*/ |
|
205
|
|
|
private function extractTransUnitMetadata( DOMElement $transUnit, array &$transUnitIdArrayForUniquenessCheck ) { |
|
206
|
|
|
$metadata = []; |
|
207
|
|
|
|
|
208
|
|
|
// id MUST NOT be null |
|
209
|
|
|
if ( null === $transUnit->attributes->getNamedItem( 'id' ) ) { |
|
|
|
|
|
|
210
|
|
|
throw new NotFoundIdInTransUnit( 'Invalid trans-unit id found. EMPTY value', 400 ); |
|
211
|
|
|
} |
|
212
|
|
|
|
|
213
|
|
|
/** |
|
214
|
|
|
* @var string $key |
|
215
|
|
|
* @var DOMAttr $element |
|
216
|
|
|
*/ |
|
217
|
|
|
foreach ( $transUnit->attributes as $key => $element ) { |
|
218
|
|
|
|
|
219
|
|
|
if ( $element->nodeName === "id" ) { |
|
220
|
|
|
|
|
221
|
|
|
$id = $element->nodeValue; |
|
222
|
|
|
|
|
223
|
|
|
if ( strlen( $id ) > 100 ) { |
|
224
|
|
|
throw new SegmentIdTooLongException( 'Segment-id too long. Max 100 characters allowed', 400 ); |
|
225
|
|
|
} |
|
226
|
|
|
|
|
227
|
|
|
$transUnitIdArrayForUniquenessCheck[] = $id; |
|
228
|
|
|
$metadata[ 'id' ] = $id; |
|
229
|
|
|
|
|
230
|
|
|
} elseif ( $element->nodeName === "approved" ) { |
|
231
|
|
|
// approved as BOOLEAN |
|
232
|
|
|
// http://docs.oasis-open.org/xliff/v1.2/os/xliff-core.html#approved |
|
233
|
|
|
$metadata[ $element->nodeName ] = filter_var( $element->nodeValue, FILTER_VALIDATE_BOOLEAN ); |
|
234
|
|
|
} elseif ( $element->nodeName === "maxwidth" ) { |
|
235
|
|
|
// we ignore ( but we get ) the attribute size-unit="char" assuming that a restriction is everytime done by character |
|
236
|
|
|
// we duplicate the info to allow Xliff V1 and V2 to work the same |
|
237
|
|
|
$metadata[ 'sizeRestriction' ] = filter_var( $element->nodeValue, FILTER_SANITIZE_NUMBER_INT ); |
|
238
|
|
|
$metadata[ $element->nodeName ] = filter_var( $element->nodeValue, FILTER_SANITIZE_NUMBER_INT ); |
|
239
|
|
|
} else { |
|
240
|
|
|
$metadata[ $element->nodeName ] = $element->nodeValue; |
|
241
|
|
|
} |
|
242
|
|
|
|
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
|
|
return $metadata; |
|
246
|
|
|
} |
|
247
|
|
|
|
|
248
|
|
|
/** |
|
249
|
|
|
* @param DOMElement $transUnit |
|
250
|
|
|
* |
|
251
|
|
|
* @return array |
|
252
|
|
|
* @throws Exception |
|
253
|
|
|
*/ |
|
254
|
|
|
private function extractTransUnitNotes( DOMDocument $dom, DOMElement $transUnit ) { |
|
255
|
|
|
$notes = []; |
|
256
|
|
|
foreach ( $transUnit->getElementsByTagName( 'note' ) as $note ) { |
|
257
|
|
|
|
|
258
|
|
|
$noteValue = $this->extractTagContent( $dom, $note ); |
|
259
|
|
|
|
|
260
|
|
|
if ( '' !== $noteValue ) { |
|
261
|
|
|
|
|
262
|
|
|
$extractedNote = $this->JSONOrRawContentArray( $noteValue ); |
|
263
|
|
|
|
|
264
|
|
|
// extract all the attributes |
|
265
|
|
|
foreach ( $note->attributes as $attribute ) { |
|
266
|
|
|
$extractedNote[ $attribute->name ] = $attribute->value; |
|
267
|
|
|
} |
|
268
|
|
|
|
|
269
|
|
|
$notes[] = $extractedNote; |
|
270
|
|
|
} |
|
271
|
|
|
} |
|
272
|
|
|
|
|
273
|
|
|
return $notes; |
|
274
|
|
|
} |
|
275
|
|
|
|
|
276
|
|
|
/** |
|
277
|
|
|
* @param DOMElement $contextGroup |
|
278
|
|
|
* |
|
279
|
|
|
* @return array |
|
280
|
|
|
*/ |
|
281
|
|
|
private function extractTransUnitContextGroup( DOMDocument $dom, DOMElement $contextGroup ) { |
|
282
|
|
|
$cg = []; |
|
283
|
|
|
$cg[ 'attr' ] = $this->extractTagAttributes( $contextGroup ); |
|
284
|
|
|
|
|
285
|
|
|
/** @var DOMNode $context */ |
|
286
|
|
|
foreach ( $contextGroup->childNodes as $context ) { |
|
287
|
|
|
if ( $context->nodeName === 'context' ) { |
|
288
|
|
|
$cg[ 'contexts' ][] = $this->extractContent( $dom, $context ); |
|
289
|
|
|
} |
|
290
|
|
|
} |
|
291
|
|
|
|
|
292
|
|
|
return $cg; |
|
293
|
|
|
} |
|
294
|
|
|
|
|
295
|
|
|
/** |
|
296
|
|
|
* @param DOMElement $altTrans |
|
297
|
|
|
* |
|
298
|
|
|
* @return array |
|
299
|
|
|
*/ |
|
300
|
|
|
private function extractTransUnitAltTrans( DOMElement $altTrans ) { |
|
301
|
|
|
$at = []; |
|
302
|
|
|
$at[ 'attr' ] = $this->extractTagAttributes( $altTrans ); |
|
303
|
|
|
|
|
304
|
|
|
if ( $altTrans->getElementsByTagName( 'source' )->length > 0 ) { |
|
305
|
|
|
$at[ 'source' ] = $altTrans->getElementsByTagName( 'source' )->item( 0 )->nodeValue; |
|
306
|
|
|
} |
|
307
|
|
|
|
|
308
|
|
|
if ( $altTrans->getElementsByTagName( 'target' ) ) { |
|
309
|
|
|
$at[ 'target' ] = $altTrans->getElementsByTagName( 'target' )->item( 0 )->nodeValue; |
|
310
|
|
|
} |
|
311
|
|
|
|
|
312
|
|
|
return $at; |
|
313
|
|
|
} |
|
314
|
|
|
|
|
315
|
|
|
/** |
|
316
|
|
|
* @param DOMElement $locked |
|
317
|
|
|
* |
|
318
|
|
|
* @return bool |
|
319
|
|
|
*/ |
|
320
|
|
|
private function extractLocked( DOMElement $locked ) { |
|
321
|
|
|
return null !== $locked->getAttribute( 'locked' ); |
|
322
|
|
|
} |
|
323
|
|
|
} |
|
324
|
|
|
|
matecat /
xliff-parser
This check looks for calls to methods that do not seem to exist on a given type. It looks for the method on the type itself as well as in inherited classes or implemented interfaces.
This is most likely a typographical error or the method has been renamed.