|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Created by PhpStorm. |
|
4
|
|
|
* @author hashashiyyin [email protected] / [email protected] |
|
5
|
|
|
* Date: 22/04/24 |
|
6
|
|
|
* Time: 15:13 |
|
7
|
|
|
* |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
namespace Matecat\SubFiltering\Utils; |
|
11
|
|
|
|
|
12
|
|
|
use DOMException; |
|
13
|
|
|
use Exception; |
|
14
|
|
|
use Matecat\SubFiltering\Enum\CTypeEnum; |
|
15
|
|
|
use Matecat\XmlParser\Exception\InvalidXmlException; |
|
16
|
|
|
use Matecat\XmlParser\Exception\XmlParsingException; |
|
17
|
|
|
use Matecat\XmlParser\XmlParser; |
|
18
|
|
|
|
|
19
|
|
|
class DataRefReplacer { |
|
20
|
|
|
/** |
|
21
|
|
|
* @var Map |
|
22
|
|
|
*/ |
|
23
|
|
|
private $map; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* DataRefReplacer constructor. |
|
27
|
|
|
* |
|
28
|
|
|
* @param array $map |
|
29
|
|
|
*/ |
|
30
|
|
|
public function __construct( array $map = [] ) { |
|
31
|
|
|
$this->map = Map::instance( $this->sanitizeMap( $map ) ); |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* This function inserts a new attribute called 'equiv-text' from dataRef contained in <ph>, <sc>, <ec>, <pc> tags against the provided map array |
|
36
|
|
|
* |
|
37
|
|
|
* For a complete reference see: |
|
38
|
|
|
* |
|
39
|
|
|
* Http://docs.oasis-open.org/xliff/xliff-core/v2.1/os/xliff-core-v2.1-os.html#dataref |
|
40
|
|
|
* |
|
41
|
|
|
* @param string $string |
|
42
|
|
|
* |
|
43
|
|
|
* @return string |
|
44
|
|
|
*/ |
|
45
|
|
|
public function replace( $string ) { |
|
46
|
|
|
|
|
47
|
|
|
// if the map is empty |
|
48
|
|
|
// or the string has not a dataRef attribute |
|
49
|
|
|
// return string as is |
|
50
|
|
|
if ( $this->map->isEmpty() || !$this->hasAnyDataRefAttribute( $string ) ) { |
|
51
|
|
|
return $string; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
// try not to throw exception for wrong segments with opening tags and no closing |
|
55
|
|
|
try { |
|
56
|
|
|
|
|
57
|
|
|
$html = XmlParser::parse( $string, true ); |
|
58
|
|
|
|
|
59
|
|
|
$dataRefEndMap = new ArrayList(); |
|
60
|
|
|
|
|
61
|
|
|
foreach ( $html as $node ) { |
|
62
|
|
|
|
|
63
|
|
|
// 1. Replace <ph>|<sc>|<ec> tags |
|
64
|
|
|
$string = $this->recursiveTransformDataRefToPhTag( $node, $string ); |
|
65
|
|
|
|
|
66
|
|
|
// 2. Replace self-closed <pc dataRefStart="xyz" /> tags |
|
67
|
|
|
$string = $this->recursiveReplaceSelfClosedPcTags( $node, $string ); |
|
68
|
|
|
|
|
69
|
|
|
// 3. Build the DataRefEndMap needed by replaceClosingPcTags function |
|
70
|
|
|
// (needed for correct handling of </pc> closing tags) |
|
71
|
|
|
// make this inline with one foreach cycle |
|
72
|
|
|
$this->extractDataRefMapRecursively( $node, $dataRefEndMap ); |
|
73
|
|
|
|
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
// 4. replace pc tags |
|
77
|
|
|
$string = $this->replaceOpeningPcTags( $string ); |
|
78
|
|
|
$string = $this->replaceClosingPcTags( $string, $dataRefEndMap ); |
|
79
|
|
|
|
|
80
|
|
|
} catch ( Exception $ignore ) { |
|
81
|
|
|
// if something fails here, do not throw exception and return the original string instead |
|
82
|
|
|
// var_dump( $ignore ); |
|
83
|
|
|
} finally { |
|
84
|
|
|
return $string; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* @param string $string |
|
91
|
|
|
* |
|
92
|
|
|
* @return bool |
|
93
|
|
|
*/ |
|
94
|
|
|
private function hasAnyDataRefAttribute( $string ) { |
|
95
|
|
|
return (bool)preg_match( '/(dataRef|dataRefStart|dataRefEnd)=[\'"].*?[\'"]/', $string ); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* This function adds equiv-text attribute to <ph>, <ec>, and <sc> tags. |
|
100
|
|
|
* |
|
101
|
|
|
* Please note that <ec> and <sc> tags are converted to <ph> tags (needed by Matecat); |
|
102
|
|
|
* in this case, another special attribute (dataType) is added just before equiv-text |
|
103
|
|
|
* |
|
104
|
|
|
* If there is no id tag, it will be copied from dataRef attribute |
|
105
|
|
|
* |
|
106
|
|
|
* @param object $node |
|
107
|
|
|
* @param string $string |
|
108
|
|
|
* |
|
109
|
|
|
* @return string |
|
110
|
|
|
*/ |
|
111
|
|
|
private function recursiveTransformDataRefToPhTag( $node, $string ) { |
|
112
|
|
|
|
|
113
|
|
|
if ( $node->has_children ) { |
|
114
|
|
|
|
|
115
|
|
|
foreach ( $node->inner_html as $childNode ) { |
|
116
|
|
|
$string = $this->recursiveTransformDataRefToPhTag( $childNode, $string ); |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
} else { |
|
120
|
|
|
|
|
121
|
|
|
// accept only those tags |
|
122
|
|
|
switch ( $node->tagName ) { |
|
123
|
|
|
case 'ph': |
|
124
|
|
|
$ctype = CTypeEnum::PH_DATA_REF; |
|
125
|
|
|
break; |
|
126
|
|
|
case 'sc': |
|
127
|
|
|
$ctype = CTypeEnum::SC_DATA_REF; |
|
128
|
|
|
break; |
|
129
|
|
|
case 'ec': |
|
130
|
|
|
$ctype = CTypeEnum::EC_DATA_REF; |
|
131
|
|
|
break; |
|
132
|
|
|
default: |
|
133
|
|
|
return $string; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
// if isset a value in the map proceed with conversion otherwise skip |
|
137
|
|
|
$attributesMap = Map::instance( $node->attributes ); |
|
138
|
|
|
if ( !$this->map->get( $attributesMap->get( 'dataRef' ) ) ) { |
|
139
|
|
|
return $string; |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
$dataRefName = $node->attributes[ 'dataRef' ]; // map identifier. Eg: source1 |
|
143
|
|
|
|
|
144
|
|
|
return $this->replaceNewTagString( |
|
145
|
|
|
$node->node, |
|
146
|
|
|
$attributesMap->getOrDefault( 'id', $dataRefName ), |
|
147
|
|
|
$this->map->getOrDefault( $node->attributes[ 'dataRef' ], 'NULL' ), |
|
148
|
|
|
$ctype, |
|
149
|
|
|
$string, |
|
150
|
|
|
null |
|
151
|
|
|
); |
|
152
|
|
|
|
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
|
|
return $string; |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
|
|
/** |
|
159
|
|
|
* Check if values in the map are null or an empty string, in that case, convert them to NULL string |
|
160
|
|
|
* |
|
161
|
|
|
* @param $map |
|
162
|
|
|
* |
|
163
|
|
|
* @return array |
|
164
|
|
|
*/ |
|
165
|
|
|
private function sanitizeMap( $map ) { |
|
166
|
|
|
|
|
167
|
|
|
foreach ( $map as $name => $value ) { |
|
168
|
|
|
if ( is_null( $value ) || $value === '' ) { |
|
169
|
|
|
$map[ $name ] = 'NULL'; |
|
170
|
|
|
} |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
|
|
return $map; |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
/** |
|
177
|
|
|
* @param $node |
|
178
|
|
|
* @param $string |
|
179
|
|
|
* |
|
180
|
|
|
* @return string |
|
181
|
|
|
* @throws DOMException |
|
182
|
|
|
* @throws InvalidXmlException |
|
183
|
|
|
* @throws XmlParsingException |
|
184
|
|
|
*/ |
|
185
|
|
|
private function recursiveReplaceSelfClosedPcTags( $node, $string ) { |
|
186
|
|
|
|
|
187
|
|
|
if ( $node->has_children ) { |
|
188
|
|
|
|
|
189
|
|
|
foreach ( $node->inner_html as $childNode ) { |
|
190
|
|
|
$string = $this->recursiveReplaceSelfClosedPcTags( $childNode, $string ); |
|
191
|
|
|
} |
|
192
|
|
|
|
|
193
|
|
|
} elseif ( $node->tagName == 'pc' && $node->self_closed === true ) { |
|
194
|
|
|
|
|
195
|
|
|
$attributesMap = Map::instance( $node->attributes ); |
|
196
|
|
|
|
|
197
|
|
|
if ( $dataRefStartValue = $this->map->get( $node->attributes[ 'dataRefStart' ] ) ) { |
|
198
|
|
|
|
|
199
|
|
|
$string = $this->replaceNewTagString( |
|
200
|
|
|
$node->node, |
|
201
|
|
|
$attributesMap->get( 'id' ), |
|
202
|
|
|
$dataRefStartValue, |
|
203
|
|
|
CTypeEnum::PC_SELF_CLOSE_DATA_REF, |
|
204
|
|
|
$string |
|
205
|
|
|
); |
|
206
|
|
|
|
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
} |
|
210
|
|
|
|
|
211
|
|
|
return $string; |
|
212
|
|
|
|
|
213
|
|
|
} |
|
214
|
|
|
|
|
215
|
|
|
/** |
|
216
|
|
|
* Extract (recursively) the dataRefEnd map from single nodes |
|
217
|
|
|
* |
|
218
|
|
|
* @param object $node |
|
219
|
|
|
* @param ArrayList $dataRefEndMap |
|
220
|
|
|
*/ |
|
221
|
|
|
private function extractDataRefMapRecursively( $node, ArrayList $dataRefEndMap ) { |
|
222
|
|
|
|
|
223
|
|
|
// we have to build the map for the closing pc tag, so get the children first |
|
224
|
|
|
if ( $node->has_children ) { |
|
225
|
|
|
foreach ( $node->inner_html as $nestedNode ) { |
|
226
|
|
|
$this->extractDataRefMapRecursively( $nestedNode, $dataRefEndMap ); |
|
227
|
|
|
} |
|
228
|
|
|
} |
|
229
|
|
|
|
|
230
|
|
|
// EXCLUDE self closed <pc/> |
|
231
|
|
|
if ( $node->tagName === 'pc' && $node->self_closed === false ) { |
|
232
|
|
|
|
|
233
|
|
|
$attributesMap = Map::instance( $node->attributes ); |
|
234
|
|
|
$dataRefEnd = $attributesMap->getOrDefault( 'dataRefEnd', $attributesMap->get( 'dataRefStart' ) ); |
|
235
|
|
|
|
|
236
|
|
|
$dataRefEndMap[] = [ |
|
237
|
|
|
'id' => $attributesMap->get( 'id' ), |
|
238
|
|
|
'dataRefEnd' => $dataRefEnd, |
|
239
|
|
|
]; |
|
240
|
|
|
|
|
241
|
|
|
} |
|
242
|
|
|
|
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
|
|
/** |
|
246
|
|
|
* Replace opening <pc> tags with correct reference in the $string |
|
247
|
|
|
* |
|
248
|
|
|
* @param string $string |
|
249
|
|
|
* |
|
250
|
|
|
* @return string |
|
251
|
|
|
* @throws DOMException |
|
252
|
|
|
* @throws InvalidXmlException |
|
253
|
|
|
* @throws XmlParsingException |
|
254
|
|
|
*/ |
|
255
|
|
|
private function replaceOpeningPcTags( $string ) { |
|
256
|
|
|
|
|
257
|
|
|
preg_match_all( '|<pc ([^>/]+?)>|iu', $string, $openingPcMatches ); |
|
258
|
|
|
|
|
259
|
|
|
foreach ( $openingPcMatches[ 0 ] as $match ) { |
|
260
|
|
|
|
|
261
|
|
|
$node = XmlParser::parse( $match . '</pc>', true )[ 0 ]; // add a closing tag to not break xml integrity |
|
262
|
|
|
|
|
263
|
|
|
// CASE 1 - Missing `dataRefStart` |
|
264
|
|
|
if ( isset( $node->attributes[ 'dataRefEnd' ] ) && !isset( $node->attributes[ 'dataRefStart' ] ) ) { |
|
265
|
|
|
$node->attributes[ 'dataRefStart' ] = $node->attributes[ 'dataRefEnd' ]; |
|
266
|
|
|
} |
|
267
|
|
|
|
|
268
|
|
|
// CASE 2 - Missing `dataRefEnd` |
|
269
|
|
|
if ( isset( $node->attributes[ 'dataRefStart' ] ) && !isset( $node->attributes[ 'dataRefEnd' ] ) ) { |
|
270
|
|
|
$node->attributes[ 'dataRefEnd' ] = $node->attributes[ 'dataRefStart' ]; |
|
271
|
|
|
} |
|
272
|
|
|
|
|
273
|
|
|
if ( isset( $node->attributes[ 'dataRefStart' ] ) ) { |
|
274
|
|
|
|
|
275
|
|
|
$attributesMap = Map::instance( $node->attributes ); |
|
276
|
|
|
$string = $this->replaceNewTagString( |
|
277
|
|
|
$match, |
|
278
|
|
|
$attributesMap->get( 'id' ), |
|
279
|
|
|
$this->map->getOrDefault( $node->attributes[ 'dataRefStart' ], 'NULL' ), |
|
280
|
|
|
CTypeEnum::PC_OPEN_DATA_REF, |
|
281
|
|
|
$string |
|
282
|
|
|
); |
|
283
|
|
|
|
|
284
|
|
|
} |
|
285
|
|
|
} |
|
286
|
|
|
|
|
287
|
|
|
return $string; |
|
288
|
|
|
} |
|
289
|
|
|
|
|
290
|
|
|
/** |
|
291
|
|
|
* Replace closing </pc> tags with correct reference in the $string |
|
292
|
|
|
* thanks to $dataRefEndMap |
|
293
|
|
|
* |
|
294
|
|
|
* @param string $string |
|
295
|
|
|
* @param ArrayList $dataRefEndMap |
|
296
|
|
|
* |
|
297
|
|
|
* @return string |
|
298
|
|
|
*/ |
|
299
|
|
|
private function replaceClosingPcTags( $string, ArrayList $dataRefEndMap ) { |
|
300
|
|
|
|
|
301
|
|
|
preg_match_all( '|</pc>|iu', $string, $closingPcMatches, PREG_OFFSET_CAPTURE ); |
|
302
|
|
|
$delta = 0; |
|
303
|
|
|
|
|
304
|
|
|
foreach ( $closingPcMatches[ 0 ] as $index => $match ) { |
|
305
|
|
|
|
|
306
|
|
|
$offset = $match[ 1 ]; |
|
307
|
|
|
$length = 5; // strlen of '</pc>' |
|
308
|
|
|
|
|
309
|
|
|
$attr = $dataRefEndMap->get( $index ); |
|
310
|
|
|
if ( !empty( $attr ) && isset( $attr[ 'dataRefEnd' ] ) ) { |
|
311
|
|
|
|
|
312
|
|
|
// conversion for opening <pc> tag |
|
313
|
|
|
$completeTag = $this->getNewTagString( |
|
314
|
|
|
'</pc>', |
|
315
|
|
|
$attr[ 'id' ], |
|
316
|
|
|
$this->map->getOrDefault( $attr[ 'dataRefEnd' ], 'NULL' ), |
|
317
|
|
|
CTypeEnum::PC_CLOSE_DATA_REF, |
|
318
|
|
|
'_2' |
|
319
|
|
|
); |
|
320
|
|
|
|
|
321
|
|
|
$realOffset = ( $delta === 0 ) ? $offset : ( $offset + $delta ); |
|
322
|
|
|
$string = substr_replace( $string, $completeTag, $realOffset, $length ); |
|
323
|
|
|
$delta = $delta + strlen( $completeTag ) - $length; |
|
324
|
|
|
|
|
325
|
|
|
} |
|
326
|
|
|
|
|
327
|
|
|
} |
|
328
|
|
|
|
|
329
|
|
|
return $string; |
|
|
|
|
|
|
330
|
|
|
|
|
331
|
|
|
} |
|
332
|
|
|
|
|
333
|
|
|
/** |
|
334
|
|
|
* @param string $string |
|
335
|
|
|
* |
|
336
|
|
|
* @return string |
|
337
|
|
|
* @throws DOMException |
|
338
|
|
|
* @throws InvalidXmlException |
|
339
|
|
|
* @throws XmlParsingException |
|
340
|
|
|
*/ |
|
341
|
|
|
public function restore( $string ) { |
|
342
|
|
|
|
|
343
|
|
|
// if the map is empty return string as is |
|
344
|
|
|
if ( empty( $this->map ) ) { |
|
345
|
|
|
return $string; |
|
346
|
|
|
} |
|
347
|
|
|
|
|
348
|
|
|
$html = XmlParser::parse( $string, true ); |
|
349
|
|
|
|
|
350
|
|
|
foreach ( $html as $node ) { |
|
351
|
|
|
$string = $this->recursiveRestoreOriginalTags( $node, $string ); |
|
352
|
|
|
} |
|
353
|
|
|
|
|
354
|
|
|
return $string; |
|
355
|
|
|
} |
|
356
|
|
|
|
|
357
|
|
|
/** |
|
358
|
|
|
* @param object $node |
|
359
|
|
|
* @param $string |
|
360
|
|
|
* |
|
361
|
|
|
* @return string |
|
362
|
|
|
*/ |
|
363
|
|
|
private function recursiveRestoreOriginalTags( $node, $string ) { |
|
364
|
|
|
|
|
365
|
|
|
if ( $node->has_children ) { |
|
366
|
|
|
|
|
367
|
|
|
foreach ( $node->inner_html as $childNode ) { |
|
368
|
|
|
$string = $this->recursiveRestoreOriginalTags( $childNode, $string ); |
|
369
|
|
|
} |
|
370
|
|
|
|
|
371
|
|
|
} else { |
|
372
|
|
|
|
|
373
|
|
|
$nodeAttributesMap = Map::instance( $node->attributes ); |
|
374
|
|
|
$cType = $nodeAttributesMap->get( 'ctype' ); |
|
375
|
|
|
|
|
376
|
|
|
if ( CTypeEnum::isLayer2Constant( $cType ) ) { |
|
377
|
|
|
return preg_replace( '/' . preg_quote( $node->node, '/' ) . '/', base64_decode( $node->attributes[ 'x-orig' ] ), $string, 1 ); |
|
378
|
|
|
} |
|
379
|
|
|
|
|
380
|
|
|
} |
|
381
|
|
|
|
|
382
|
|
|
return $string; |
|
383
|
|
|
|
|
384
|
|
|
} |
|
385
|
|
|
|
|
386
|
|
|
/** |
|
387
|
|
|
* @param string $actualNodeString |
|
388
|
|
|
* @param string $id |
|
389
|
|
|
* @param string $dataRefValue |
|
390
|
|
|
* @param string $ctype |
|
391
|
|
|
* @param string|null $upCountIdValue |
|
392
|
|
|
* |
|
393
|
|
|
* @return string |
|
394
|
|
|
*/ |
|
395
|
|
|
private function getNewTagString( $actualNodeString, $id, $dataRefValue, $ctype, $upCountIdValue = null ) { |
|
396
|
|
|
|
|
397
|
|
|
$newTag = [ '<ph' ]; |
|
398
|
|
|
|
|
399
|
|
|
if ( isset( $id ) ) { |
|
400
|
|
|
$newTag[] = 'id="' . $id . $upCountIdValue . '"'; |
|
401
|
|
|
} |
|
402
|
|
|
|
|
403
|
|
|
$newTag[] = 'ctype="' . $ctype . '"'; |
|
404
|
|
|
$newTag[] = 'equiv-text="base64:' . base64_encode( $dataRefValue ) . '"'; |
|
405
|
|
|
$newTag[] = 'x-orig="' . base64_encode( $actualNodeString ) . '"'; |
|
406
|
|
|
|
|
407
|
|
|
return implode( " ", $newTag ) . '/>'; |
|
408
|
|
|
|
|
409
|
|
|
} |
|
410
|
|
|
|
|
411
|
|
|
private function replaceNewTagString( $actualNodeString, $id, $dataRefValue, $ctype, $originalString, $upCountIdValue = '_1' ) { |
|
412
|
|
|
return str_replace( $actualNodeString, $this->getNewTagString( $actualNodeString, $id, $dataRefValue, $ctype, $upCountIdValue ), $originalString ); |
|
413
|
|
|
} |
|
414
|
|
|
|
|
415
|
|
|
} |