|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Created by PhpStorm. |
|
4
|
|
|
* @author hashashiyyin [email protected] / [email protected] |
|
5
|
|
|
* Date: 02/08/24 |
|
6
|
|
|
* Time: 11:45 |
|
7
|
|
|
* |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
namespace Matecat\XliffParser\XliffReplacer; |
|
11
|
|
|
|
|
12
|
|
|
use Matecat\XliffParser\Utils\Strings; |
|
13
|
|
|
|
|
14
|
|
|
class Xliff12 extends AbstractXliffReplacer { |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* @var array |
|
18
|
|
|
*/ |
|
19
|
|
|
protected array $nodesToBuffer = [ |
|
20
|
|
|
'source', |
|
21
|
|
|
'seg-source', |
|
22
|
|
|
'note', |
|
23
|
|
|
'context-group' |
|
24
|
|
|
]; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @var string |
|
28
|
|
|
*/ |
|
29
|
|
|
protected string $tuTagName = 'trans-unit'; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @var string |
|
33
|
|
|
*/ |
|
34
|
|
|
protected string $alternativeMatchesTag = 'alt-trans'; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @var string |
|
38
|
|
|
*/ |
|
39
|
|
|
protected string $namespace = "mtc"; // Custom namespace |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @inheritDoc |
|
43
|
|
|
*/ |
|
44
|
|
|
protected function tagOpen( $parser, string $name, array $attr ) { |
|
45
|
|
|
|
|
46
|
|
|
$this->handleOpenUnit( $name, $attr ); |
|
47
|
|
|
|
|
48
|
|
|
$this->trySetAltTrans( $name );; |
|
49
|
|
|
$this->checkSetInTarget( $name ); |
|
50
|
|
|
|
|
51
|
|
|
// open buffer |
|
52
|
|
|
$this->setInBuffer( $name ); |
|
53
|
|
|
|
|
54
|
|
|
// check if we are inside a <target>, obviously this happen only if there are targets inside the trans-unit |
|
55
|
|
|
// <target> must be stripped to be replaced, so this check avoids <target> reconstruction |
|
56
|
|
|
if ( !$this->inTarget ) { |
|
57
|
|
|
|
|
58
|
|
|
$tag = ''; |
|
59
|
|
|
|
|
60
|
|
|
// construct tag |
|
61
|
|
|
$tag .= "<$name "; |
|
62
|
|
|
|
|
63
|
|
|
foreach ( $attr as $k => $v ) { |
|
64
|
|
|
|
|
65
|
|
|
//if tag name is file, we must replace the target-language attribute |
|
66
|
|
|
if ( $name === 'file' && $k === 'target-language' && !empty( $this->targetLang ) ) { |
|
67
|
|
|
//replace Target language with job language provided from constructor |
|
68
|
|
|
$tag .= "$k=\"$this->targetLang\" "; |
|
69
|
|
|
} else { |
|
70
|
|
|
$tag .= "$k=\"$v\" "; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
$seg = $this->getCurrentSegment(); |
|
76
|
|
|
|
|
77
|
|
|
if ( $name === $this->tuTagName && !empty( $seg ) && isset( $seg[ 'sid' ] ) ) { |
|
78
|
|
|
|
|
79
|
|
|
// add `help-id` to xliff v.1* |
|
80
|
|
|
if ( strpos( $tag, 'help-id' ) === false ) { |
|
81
|
|
|
if ( !empty( $seg[ 'sid' ] ) ) { |
|
82
|
|
|
$tag .= "help-id=\"{$seg[ 'sid' ]}\" "; |
|
83
|
|
|
} |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
$tag = $this->handleOpenXliffTag( $name, $attr, $tag ); |
|
89
|
|
|
|
|
90
|
|
|
$this->checkForSelfClosedTagAndFlush( $parser, $tag ); |
|
91
|
|
|
|
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
|
|
97
|
|
|
/** |
|
98
|
|
|
* @inheritDoc |
|
99
|
|
|
*/ |
|
100
|
|
|
protected function tagClose( $parser, string $name ) { |
|
101
|
|
|
$tag = ''; |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* if is a tag within <target> or |
|
105
|
|
|
* if it is an empty tag, do not add closing tag because we have already closed it in |
|
106
|
|
|
* |
|
107
|
|
|
* self::tagOpen method |
|
108
|
|
|
*/ |
|
109
|
|
|
if ( !$this->isEmpty ) { |
|
110
|
|
|
|
|
111
|
|
|
if ( !$this->inTarget ) { |
|
112
|
|
|
$tag = "</$name>"; |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
if ( 'target' == $name && !$this->inAltTrans ) { |
|
116
|
|
|
|
|
117
|
|
|
if ( isset( $this->transUnits[ $this->currentTransUnitId ] ) ) { |
|
118
|
|
|
|
|
119
|
|
|
// get translation of current segment, by indirect indexing: id -> positional index -> segment |
|
120
|
|
|
// actually there may be more than one segment to that ID if there are two mrk of the same source segment |
|
121
|
|
|
$tag = $this->rebuildTarget(); |
|
122
|
|
|
|
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
$this->targetWasWritten = true; |
|
126
|
|
|
// signal we are leaving a target |
|
127
|
|
|
$this->inTarget = false; |
|
128
|
|
|
$this->postProcAndFlush( $this->outputFP, $tag, true ); |
|
129
|
|
|
|
|
130
|
|
|
} elseif ( in_array( $name, $this->nodesToBuffer ) ) { // we are closing a critical CDATA section |
|
131
|
|
|
|
|
132
|
|
|
$this->bufferIsActive = false; |
|
133
|
|
|
$tag = $this->CDATABuffer . "</$name>"; |
|
134
|
|
|
$this->CDATABuffer = ""; |
|
135
|
|
|
|
|
136
|
|
|
//flush to the pointer |
|
137
|
|
|
$this->postProcAndFlush( $this->outputFP, $tag ); |
|
138
|
|
|
|
|
139
|
|
|
} elseif ( $name === $this->tuTagName ) { |
|
140
|
|
|
|
|
141
|
|
|
$tag = ""; |
|
142
|
|
|
|
|
143
|
|
|
// handling </trans-unit> closure |
|
144
|
|
|
if ( !$this->targetWasWritten ) { |
|
145
|
|
|
|
|
146
|
|
|
if ( isset( $this->transUnits[ $this->currentTransUnitId ] ) ) { |
|
147
|
|
|
$tag = $this->rebuildTarget(); |
|
148
|
|
|
} else { |
|
149
|
|
|
$tag = $this->createTargetTag( "", "" ); |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
} |
|
153
|
|
|
|
|
154
|
|
|
$tag .= "</$this->tuTagName>"; |
|
155
|
|
|
$this->targetWasWritten = false; |
|
156
|
|
|
$this->postProcAndFlush( $this->outputFP, $tag ); |
|
157
|
|
|
|
|
158
|
|
|
} elseif ( $this->bufferIsActive ) { // this is a tag ( <g | <mrk ) inside a seg or seg-source tag |
|
159
|
|
|
$this->CDATABuffer .= "</$name>"; |
|
160
|
|
|
// Do NOT Flush |
|
161
|
|
|
} else { //generic tag closure do Nothing |
|
162
|
|
|
// flush to pointer |
|
163
|
|
|
$this->postProcAndFlush( $this->outputFP, $tag ); |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
|
|
} else { |
|
167
|
|
|
//ok, nothing to be done; reset flag for next coming tag |
|
168
|
|
|
$this->isEmpty = false; |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
|
|
// try to signal that we are leaving a target |
|
172
|
|
|
$this->tryUnsetAltTrans( $name ); |
|
173
|
|
|
|
|
174
|
|
|
// check if we are leaving a <trans-unit> (xliff v1.*) or <unit> (xliff v2.*) |
|
175
|
|
|
if ( $this->tuTagName === $name ) { |
|
176
|
|
|
$this->currentTransUnitIsTranslatable = null; |
|
177
|
|
|
$this->inTU = false; |
|
178
|
|
|
$this->hasWrittenCounts = false; |
|
179
|
|
|
|
|
180
|
|
|
$this->resetCounts(); |
|
181
|
|
|
} |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
/** |
|
185
|
|
|
* prepare segment tagging for xliff insertion |
|
186
|
|
|
* |
|
187
|
|
|
* @param array $seg |
|
188
|
|
|
* @param string $transUnitTranslation |
|
189
|
|
|
* |
|
190
|
|
|
* @return string |
|
191
|
|
|
*/ |
|
192
|
|
|
protected function prepareTranslation( array $seg, string $transUnitTranslation = "" ): string { |
|
193
|
|
|
|
|
194
|
|
|
$segment = Strings::removeDangerousChars( $seg [ 'segment' ] ); |
|
195
|
|
|
$translation = Strings::removeDangerousChars( $seg [ 'translation' ] ); |
|
196
|
|
|
|
|
197
|
|
|
if ( $seg [ 'translation' ] == '' ) { |
|
198
|
|
|
$translation = $segment; |
|
199
|
|
|
} else { |
|
200
|
|
|
if ( $this->callback instanceof XliffReplacerCallbackInterface ) { |
|
201
|
|
|
$error = ( !empty( $seg[ 'error' ] ) ) ? $seg[ 'error' ] : null; |
|
202
|
|
|
if ( $this->callback->thereAreErrors( $seg[ 'sid' ], $segment, $translation, [], $error ) ) { |
|
203
|
|
|
$translation = '|||UNTRANSLATED_CONTENT_START|||' . $segment . '|||UNTRANSLATED_CONTENT_END|||'; |
|
204
|
|
|
} |
|
205
|
|
|
} |
|
206
|
|
|
} |
|
207
|
|
|
|
|
208
|
|
|
$transUnitTranslation .= $seg[ 'prev_tags' ] . $this->rebuildMarks( $seg, $translation ) . ltrim( $seg[ 'succ_tags' ] ); |
|
209
|
|
|
|
|
210
|
|
|
return $transUnitTranslation; |
|
211
|
|
|
} |
|
212
|
|
|
|
|
213
|
|
|
protected function rebuildMarks( array $seg, string $translation ): string { |
|
214
|
|
|
|
|
215
|
|
|
if ( $seg[ 'mrk_id' ] !== null && $seg[ 'mrk_id' ] != '' ) { |
|
216
|
|
|
$translation = "<mrk mid=\"" . $seg[ 'mrk_id' ] . "\" mtype=\"seg\">" . $seg[ 'mrk_prev_tags' ] . $translation . $seg[ 'mrk_succ_tags' ] . "</mrk>"; |
|
217
|
|
|
} |
|
218
|
|
|
|
|
219
|
|
|
return $translation; |
|
220
|
|
|
|
|
221
|
|
|
} |
|
222
|
|
|
|
|
223
|
|
|
/** |
|
224
|
|
|
* This function creates a <target> |
|
225
|
|
|
* |
|
226
|
|
|
* @param string $translation |
|
227
|
|
|
* @param string $stateProp |
|
228
|
|
|
* |
|
229
|
|
|
* @return string |
|
230
|
|
|
*/ |
|
231
|
|
|
private function createTargetTag( string $translation, string $stateProp ): string { |
|
232
|
|
|
$targetLang = ' xml:lang="' . $this->targetLang . '"'; |
|
233
|
|
|
$tag = "<target $targetLang $stateProp>$translation</target>"; |
|
234
|
|
|
$tag .= "\n<count-group name=\"$this->currentTransUnitId\"><count count-type=\"x-matecat-raw\">" . $this->counts[ 'raw_word_count' ] . "</count><count count-type=\"x-matecat-weighted\">" . $this->counts[ 'eq_word_count' ] . '</count></count-group>'; |
|
235
|
|
|
|
|
236
|
|
|
return $tag; |
|
237
|
|
|
|
|
238
|
|
|
} |
|
239
|
|
|
|
|
240
|
|
|
protected function rebuildTarget(): string { |
|
241
|
|
|
|
|
242
|
|
|
// init translation and state |
|
243
|
|
|
$translation = ''; |
|
244
|
|
|
$lastMrkState = null; |
|
245
|
|
|
$stateProp = ''; |
|
246
|
|
|
|
|
247
|
|
|
// we must reset the lastMrkId found because this is a new segment. |
|
248
|
|
|
$lastMrkId = -1; |
|
249
|
|
|
|
|
250
|
|
|
foreach ( $this->lastTransUnit as $pos => $seg ) { |
|
251
|
|
|
|
|
252
|
|
|
/* |
|
253
|
|
|
* This routine works to respect the positional orders of markers. |
|
254
|
|
|
* In every cycle we check if the mrk of the segment is below or equal the last one. |
|
255
|
|
|
* When this is true, means that the mrk id belongs to the next segment with the same internal_id |
|
256
|
|
|
* so we MUST stop to apply markers and translations |
|
257
|
|
|
* and stop to add eq_word_count |
|
258
|
|
|
* |
|
259
|
|
|
* Begin: |
|
260
|
|
|
* pre-assign zero to the new mrk if this is the first one ( in this segment ) |
|
261
|
|
|
* If it is null leave it NULL |
|
262
|
|
|
*/ |
|
263
|
|
|
if ( (int)$seg[ "mrk_id" ] < 0 && $seg[ "mrk_id" ] !== null ) { |
|
264
|
|
|
$seg[ "mrk_id" ] = 0; |
|
265
|
|
|
} |
|
266
|
|
|
|
|
267
|
|
|
/* |
|
268
|
|
|
* WARNING: |
|
269
|
|
|
* For those seg-source that doesn't have a mrk ( having a mrk id === null ) |
|
270
|
|
|
* ( null <= -1 ) === true |
|
271
|
|
|
* so, cast to int |
|
272
|
|
|
*/ |
|
273
|
|
|
if ( (int)$seg[ "mrk_id" ] <= $lastMrkId ) { |
|
274
|
|
|
break; |
|
275
|
|
|
} |
|
276
|
|
|
|
|
277
|
|
|
// update counts |
|
278
|
|
|
if ( !empty( $seg ) ) { |
|
279
|
|
|
$this->updateSegmentCounts( $seg ); |
|
280
|
|
|
} |
|
281
|
|
|
|
|
282
|
|
|
// delete translations so the prepareSegment |
|
283
|
|
|
// will put source content in target tag |
|
284
|
|
|
if ( $this->sourceInTarget ) { |
|
285
|
|
|
$seg[ 'translation' ] = ''; |
|
286
|
|
|
$this->resetCounts(); |
|
287
|
|
|
} |
|
288
|
|
|
|
|
289
|
|
|
// append $translation |
|
290
|
|
|
$translation = $this->prepareTranslation( $seg, $translation ); |
|
291
|
|
|
|
|
292
|
|
|
$lastMrkId = $seg[ "mrk_id" ]; |
|
293
|
|
|
|
|
294
|
|
|
[ $stateProp, $lastMrkState ] = StatusToStateAttribute::getState( $this->xliffVersion, $seg[ 'status' ], $lastMrkState ); |
|
295
|
|
|
|
|
296
|
|
|
} |
|
297
|
|
|
|
|
298
|
|
|
//append translation |
|
299
|
|
|
return $this->createTargetTag( $translation, $stateProp ); |
|
300
|
|
|
|
|
301
|
|
|
} |
|
302
|
|
|
|
|
303
|
|
|
} |
matecat /
xliff-parser
