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