1 | // Copyright (C) 2016 The Qt Company Ltd. |
---|---|
2 | // SPDX-License-Identifier: LicenseRef-Qt-Commercial OR GPL-3.0-only WITH Qt-GPL-exception-1.0 |
3 | |
4 | #include "translator.h" |
5 | #include "xmlparser.h" |
6 | |
7 | #include <QtCore/QDebug> |
8 | #include <QtCore/QMap> |
9 | #include <QtCore/QRegularExpression> |
10 | #include <QtCore/QStack> |
11 | #include <QtCore/QString> |
12 | #include <QtCore/QTextStream> |
13 | |
14 | // The string value is historical and reflects the main purpose: Keeping |
15 | // obsolete entries separate from the magic file message (which both have |
16 | // no location information, but typically reside at opposite ends of the file). |
17 | #define MAGIC_OBSOLETE_REFERENCE "Obsolete_PO_entries" |
18 | |
19 | QT_BEGIN_NAMESPACE |
20 | |
21 | /** |
22 | * Implementation of XLIFF file format for Linguist |
23 | */ |
24 | //static const char *restypeDomain = "x-gettext-domain"; |
25 | static const char *restypeContext = "x-trolltech-linguist-context"; |
26 | static const char *restypePlurals = "x-gettext-plurals"; |
27 | static const char *restypeDummy = "x-dummy"; |
28 | static const char *dataTypeUIFile = "x-trolltech-designer-ui"; |
29 | static const char *contextMsgctxt = "x-gettext-msgctxt"; // XXX Troll invention, so far. |
30 | static const char *contextOldMsgctxt = "x-gettext-previous-msgctxt"; // XXX Troll invention, so far. |
31 | static const char *attribPlural = "trolltech:plural"; |
32 | static const char *XLIFF11namespaceURI = "urn:oasis:names:tc:xliff:document:1.1"; |
33 | static const char *XLIFF12namespaceURI = "urn:oasis:names:tc:xliff:document:1.2"; |
34 | static const char *TrollTsNamespaceURI = "urn:trolltech:names:ts:document:1.0"; |
35 | |
36 | #define COMBINE4CHARS(c1, c2, c3, c4) \ |
37 | (int(c1) << 24 | int(c2) << 16 | int(c3) << 8 | int(c4) ) |
38 | |
39 | static QString dataType(const TranslatorMessage &m) |
40 | { |
41 | QByteArray fileName = m.fileName().toLatin1(); |
42 | unsigned int extHash = 0; |
43 | int pos = fileName.size() - 1; |
44 | for (int pass = 0; pass < 4 && pos >=0; ++pass, --pos) { |
45 | if (fileName.at(i: pos) == '.') |
46 | break; |
47 | extHash |= ((int)fileName.at(i: pos) << (8*pass)); |
48 | } |
49 | |
50 | switch (extHash) { |
51 | case COMBINE4CHARS(0,'c','p','p'): |
52 | case COMBINE4CHARS(0,'c','x','x'): |
53 | case COMBINE4CHARS(0,'c','+','+'): |
54 | case COMBINE4CHARS(0,'h','p','p'): |
55 | case COMBINE4CHARS(0,'h','x','x'): |
56 | case COMBINE4CHARS(0,'h','+','+'): |
57 | return QLatin1String("cpp"); |
58 | case COMBINE4CHARS(0, 0 , 0 ,'c'): |
59 | case COMBINE4CHARS(0, 0 , 0 ,'h'): |
60 | case COMBINE4CHARS(0, 0 ,'c','c'): |
61 | case COMBINE4CHARS(0, 0 ,'c','h'): |
62 | case COMBINE4CHARS(0, 0 ,'h','h'): |
63 | return QLatin1String("c"); |
64 | case COMBINE4CHARS(0, 0 ,'u','i'): |
65 | return QLatin1String(dataTypeUIFile); //### form? |
66 | default: |
67 | return QLatin1String("plaintext"); // we give up |
68 | } |
69 | } |
70 | |
71 | static void writeIndent(QTextStream &ts, int indent) |
72 | { |
73 | ts << QString().fill(c: QLatin1Char(' '), size: indent * 2); |
74 | } |
75 | |
76 | struct CharMnemonic |
77 | { |
78 | char ch; |
79 | char escape; |
80 | const char *mnemonic; |
81 | }; |
82 | |
83 | static const CharMnemonic charCodeMnemonics[] = { |
84 | {.ch: 0x07, .escape: 'a', .mnemonic: "bel"}, |
85 | {.ch: 0x08, .escape: 'b', .mnemonic: "bs"}, |
86 | {.ch: 0x09, .escape: 't', .mnemonic: "tab"}, |
87 | {.ch: 0x0a, .escape: 'n', .mnemonic: "lf"}, |
88 | {.ch: 0x0b, .escape: 'v', .mnemonic: "vt"}, |
89 | {.ch: 0x0c, .escape: 'f', .mnemonic: "ff"}, |
90 | {.ch: 0x0d, .escape: 'r', .mnemonic: "cr"} |
91 | }; |
92 | |
93 | static char charFromEscape(char escape) |
94 | { |
95 | for (uint i = 0; i < sizeof(charCodeMnemonics)/sizeof(CharMnemonic); ++i) { |
96 | CharMnemonic cm = charCodeMnemonics[i]; |
97 | if (cm.escape == escape) |
98 | return cm.ch; |
99 | } |
100 | Q_ASSERT(0); |
101 | return escape; |
102 | } |
103 | |
104 | static QString xlNumericEntity(int ch, bool makePhs) |
105 | { |
106 | // ### This needs to be reviewed, to reflect the updated XLIFF-PO spec. |
107 | if (!makePhs || ch < 7 || ch > 0x0d) |
108 | return QString::fromLatin1(ba: "&#x%1;").arg(a: QString::number(ch, base: 16)); |
109 | |
110 | CharMnemonic cm = charCodeMnemonics[int(ch) - 7]; |
111 | QString name = QLatin1String(cm.mnemonic); |
112 | char escapechar = cm.escape; |
113 | |
114 | static int id = 0; |
115 | return QString::fromLatin1(ba: "<ph id=\"ph%1\" ctype=\"x-ch-%2\">\\%3</ph>") |
116 | .arg(a: ++id) .arg(a: name) .arg(a: escapechar); |
117 | } |
118 | |
119 | static QString xlProtect(const QString &str, bool makePhs = true) |
120 | { |
121 | QString result; |
122 | int len = str.size(); |
123 | for (int i = 0; i != len; ++i) { |
124 | uint c = str.at(i).unicode(); |
125 | switch (c) { |
126 | case '\"': |
127 | result += QLatin1String("""); |
128 | break; |
129 | case '&': |
130 | result += QLatin1String("&"); |
131 | break; |
132 | case '>': |
133 | result += QLatin1String(">"); |
134 | break; |
135 | case '<': |
136 | result += QLatin1String("<"); |
137 | break; |
138 | case '\'': |
139 | result += QLatin1String("'"); |
140 | break; |
141 | default: |
142 | if (c < 0x20 && c != '\r' && c != '\n' && c != '\t') |
143 | result += xlNumericEntity(ch: c, makePhs); |
144 | else // this also covers surrogates |
145 | result += QChar(c); |
146 | } |
147 | } |
148 | return result; |
149 | } |
150 | |
151 | |
152 | static void writeExtras(QTextStream &ts, int indent, |
153 | const TranslatorMessage::ExtraData &extras, QRegularExpression drops) |
154 | { |
155 | for (auto it = extras.cbegin(), end = extras.cend(); it != end; ++it) { |
156 | if (!drops.match(subject: it.key()).hasMatch()) { |
157 | writeIndent(ts, indent); |
158 | ts << "<trolltech:"<< it.key() << '>' |
159 | << xlProtect(str: it.value()) |
160 | << "</trolltech:"<< it.key() << ">\n"; |
161 | } |
162 | } |
163 | } |
164 | |
165 | static void writeLineNumber(QTextStream &ts, const TranslatorMessage &msg, int indent) |
166 | { |
167 | if (msg.lineNumber() == -1) |
168 | return; |
169 | writeIndent(ts, indent); |
170 | ts << "<context-group purpose=\"location\"><context context-type=\"linenumber\">" |
171 | << msg.lineNumber() << "</context></context-group>\n"; |
172 | const auto refs = msg.extraReferences(); |
173 | for (const TranslatorMessage::Reference &ref : refs) { |
174 | writeIndent(ts, indent); |
175 | ts << "<context-group purpose=\"location\">"; |
176 | if (ref.fileName() != msg.fileName()) |
177 | ts << "<context context-type=\"sourcefile\">"<< ref.fileName() << "</context>"; |
178 | ts << "<context context-type=\"linenumber\">"<< ref.lineNumber() |
179 | << "</context></context-group>\n"; |
180 | } |
181 | } |
182 | |
183 | static void writeComment(QTextStream &ts, const TranslatorMessage &msg, const QRegularExpression &drops, int indent) |
184 | { |
185 | if (!msg.comment().isEmpty()) { |
186 | writeIndent(ts, indent); |
187 | ts << "<context-group><context context-type=\""<< contextMsgctxt << "\">" |
188 | << xlProtect(str: msg.comment(), makePhs: false) |
189 | << "</context></context-group>\n"; |
190 | } |
191 | if (!msg.oldComment().isEmpty()) { |
192 | writeIndent(ts, indent); |
193 | ts << "<context-group><context context-type=\""<< contextOldMsgctxt << "\">" |
194 | << xlProtect(str: msg.oldComment(), makePhs: false) |
195 | << "</context></context-group>\n"; |
196 | } |
197 | writeExtras(ts, indent, extras: msg.extras(), drops); |
198 | if (!msg.extraComment().isEmpty()) { |
199 | writeIndent(ts, indent); |
200 | ts << "<note annotates=\"source\" from=\"developer\">" |
201 | << xlProtect(str: msg.extraComment()) << "</note>\n"; |
202 | } |
203 | if (!msg.translatorComment().isEmpty()) { |
204 | writeIndent(ts, indent); |
205 | ts << "<note from=\"translator\">" |
206 | << xlProtect(str: msg.translatorComment()) << "</note>\n"; |
207 | } |
208 | } |
209 | |
210 | static void writeTransUnits(QTextStream &ts, const TranslatorMessage &msg, const QRegularExpression &drops, int indent) |
211 | { |
212 | static int msgid; |
213 | QString msgidstr = !msg.id().isEmpty() ? msg.id() : QString::fromLatin1(ba: "_msg%1").arg(a: ++msgid); |
214 | |
215 | QStringList translns = msg.translations(); |
216 | QString pluralStr; |
217 | QStringList sources(msg.sourceText()); |
218 | const auto &extras = msg.extras(); |
219 | const auto extrasEnd = extras.cend(); |
220 | if (const auto it = extras.constFind(key: QString::fromLatin1(ba: "po-msgid_plural")); it != extrasEnd) |
221 | sources.append(t: *it); |
222 | QStringList oldsources; |
223 | if (!msg.oldSourceText().isEmpty()) |
224 | oldsources.append(t: msg.oldSourceText()); |
225 | if (const auto it = extras.constFind(key: QString::fromLatin1(ba: "po-old_msgid_plural")); it != extrasEnd) { |
226 | if (oldsources.isEmpty()) { |
227 | if (sources.size() == 2) |
228 | oldsources.append(t: QString()); |
229 | else |
230 | pluralStr = QLatin1Char(' ') + QLatin1String(attribPlural) + QLatin1String("=\"yes\""); |
231 | } |
232 | oldsources.append(t: *it); |
233 | } |
234 | |
235 | auto srcit = sources.cbegin(), srcend = sources.cend(), |
236 | oldsrcit = oldsources.cbegin(), oldsrcend = oldsources.cend(), |
237 | transit = translns.cbegin(), transend = translns.cend(); |
238 | int plural = 0; |
239 | QString source; |
240 | while (srcit != srcend || oldsrcit != oldsrcend || transit != transend) { |
241 | QByteArray attribs; |
242 | QByteArray state; |
243 | if ((msg.type() == TranslatorMessage::Obsolete |
244 | || msg.type() == TranslatorMessage::Vanished) |
245 | && !msg.isPlural()) { |
246 | attribs = " translate=\"no\""; |
247 | } |
248 | if (msg.type() == TranslatorMessage::Finished |
249 | || msg.type() == TranslatorMessage::Vanished) { |
250 | attribs += " approved=\"yes\""; |
251 | } else if (msg.type() == TranslatorMessage::Unfinished |
252 | && transit != transend && !transit->isEmpty()) { |
253 | state = " state=\"needs-review-translation\""; |
254 | } |
255 | writeIndent(ts, indent); |
256 | ts << "<trans-unit id=\""<< msgidstr; |
257 | if (msg.isPlural()) |
258 | ts << "["<< plural++ << "]"; |
259 | ts << "\""<< attribs << ">\n"; |
260 | ++indent; |
261 | |
262 | writeIndent(ts, indent); |
263 | if (srcit != srcend) { |
264 | source = *srcit; |
265 | ++srcit; |
266 | } // else just repeat last element |
267 | ts << "<source xml:space=\"preserve\">"<< xlProtect(str: source) << "</source>\n"; |
268 | |
269 | bool puttrans = false; |
270 | QString translation; |
271 | if (transit != transend) { |
272 | translation = *transit; |
273 | translation.replace(before: QChar(Translator::BinaryVariantSeparator), |
274 | after: QChar(Translator::TextVariantSeparator)); |
275 | ++transit; |
276 | puttrans = true; |
277 | } |
278 | do { |
279 | if (oldsrcit != oldsrcend && !oldsrcit->isEmpty()) { |
280 | writeIndent(ts, indent); |
281 | ts << "<alt-trans>\n"; |
282 | ++indent; |
283 | writeIndent(ts, indent); |
284 | ts << "<source xml:space=\"preserve\""<< pluralStr << '>' << xlProtect(str: *oldsrcit) << "</source>\n"; |
285 | if (!puttrans) { |
286 | writeIndent(ts, indent); |
287 | ts << "<target restype=\""<< restypeDummy << "\"/>\n"; |
288 | } |
289 | } |
290 | |
291 | if (puttrans) { |
292 | writeIndent(ts, indent); |
293 | ts << "<target xml:space=\"preserve\""<< state << ">"<< xlProtect(str: translation) << "</target>\n"; |
294 | } |
295 | |
296 | if (oldsrcit != oldsrcend) { |
297 | if (!oldsrcit->isEmpty()) { |
298 | --indent; |
299 | writeIndent(ts, indent); |
300 | ts << "</alt-trans>\n"; |
301 | } |
302 | ++oldsrcit; |
303 | } |
304 | |
305 | puttrans = false; |
306 | } while (srcit == srcend && oldsrcit != oldsrcend); |
307 | |
308 | if (!msg.isPlural()) { |
309 | writeLineNumber(ts, msg, indent); |
310 | writeComment(ts, msg, drops, indent); |
311 | } |
312 | |
313 | --indent; |
314 | writeIndent(ts, indent); |
315 | ts << "</trans-unit>\n"; |
316 | } |
317 | } |
318 | |
319 | static void writeMessage(QTextStream &ts, const TranslatorMessage &msg, const QRegularExpression &drops, int indent) |
320 | { |
321 | if (msg.isPlural()) { |
322 | writeIndent(ts, indent); |
323 | ts << "<group restype=\""<< restypePlurals << "\""; |
324 | if (!msg.id().isEmpty()) |
325 | ts << " id=\""<< msg.id() << "\""; |
326 | if (msg.type() == TranslatorMessage::Obsolete || msg.type() == TranslatorMessage::Vanished) |
327 | ts << " translate=\"no\""; |
328 | ts << ">\n"; |
329 | ++indent; |
330 | writeLineNumber(ts, msg, indent); |
331 | writeComment(ts, msg, drops, indent); |
332 | |
333 | writeTransUnits(ts, msg, drops, indent); |
334 | --indent; |
335 | writeIndent(ts, indent); |
336 | ts << "</group>\n"; |
337 | } else { |
338 | writeTransUnits(ts, msg, drops, indent); |
339 | } |
340 | } |
341 | |
342 | class XLIFFHandler : public XmlParser |
343 | { |
344 | public: |
345 | XLIFFHandler(Translator &translator, ConversionData &cd, QXmlStreamReader &reader); |
346 | ~XLIFFHandler() override = default; |
347 | |
348 | private: |
349 | bool startElement(QStringView namespaceURI, QStringView localName, |
350 | QStringView qName, const QXmlStreamAttributes &atts) override; |
351 | bool endElement(QStringView namespaceURI, QStringView localName, |
352 | QStringView qName) override; |
353 | bool characters(QStringView ch) override; |
354 | bool fatalError(qint64 line, qint64 column, const QString &message) override; |
355 | |
356 | bool endDocument() override; |
357 | |
358 | enum XliffContext { |
359 | XC_xliff, |
360 | XC_group, |
361 | XC_trans_unit, |
362 | XC_context_group, |
363 | XC_context_group_any, |
364 | XC_context, |
365 | XC_context_filename, |
366 | XC_context_linenumber, |
367 | XC_context_context, |
368 | XC_context_comment, |
369 | XC_context_old_comment, |
370 | XC_ph, |
371 | XC_extra_comment, |
372 | XC_translator_comment, |
373 | XC_restype_context, |
374 | XC_restype_translation, |
375 | XC_mtype_seg_translation, |
376 | XC_restype_plurals, |
377 | XC_alt_trans |
378 | }; |
379 | void pushContext(XliffContext ctx); |
380 | bool popContext(XliffContext ctx); |
381 | XliffContext currentContext() const; |
382 | bool hasContext(XliffContext ctx) const; |
383 | bool finalizeMessage(bool isPlural); |
384 | |
385 | private: |
386 | Translator &m_translator; |
387 | ConversionData &m_cd; |
388 | QString m_language; |
389 | QString m_sourceLanguage; |
390 | QString m_context; |
391 | QString m_id; |
392 | QStringList m_sources; |
393 | QStringList m_oldSources; |
394 | QString m_comment; |
395 | QString m_oldComment; |
396 | QString m_extraComment; |
397 | QString m_translatorComment; |
398 | bool m_translate; |
399 | bool m_approved; |
400 | bool m_isPlural; |
401 | bool m_hadAlt; |
402 | QStringList m_translations; |
403 | QString m_fileName; |
404 | int m_lineNumber; |
405 | QString m_extraFileName; |
406 | TranslatorMessage::References m_refs; |
407 | TranslatorMessage::ExtraData m_extra; |
408 | |
409 | QString accum; |
410 | QString m_ctype; |
411 | const QString m_URITT; // convenience and efficiency |
412 | const QString m_URI; // ... |
413 | const QString m_URI12; // ... |
414 | QStack<int> m_contextStack; |
415 | }; |
416 | |
417 | XLIFFHandler::XLIFFHandler(Translator &translator, ConversionData &cd, QXmlStreamReader &reader) |
418 | : XmlParser(reader, true), |
419 | m_translator(translator), |
420 | m_cd(cd), |
421 | m_translate(true), |
422 | m_approved(true), |
423 | m_lineNumber(-1), |
424 | m_URITT(QLatin1String(TrollTsNamespaceURI)), |
425 | m_URI(QLatin1String(XLIFF11namespaceURI)), |
426 | m_URI12(QLatin1String(XLIFF12namespaceURI)) |
427 | {} |
428 | |
429 | |
430 | void XLIFFHandler::pushContext(XliffContext ctx) |
431 | { |
432 | m_contextStack.push_back(t: ctx); |
433 | } |
434 | |
435 | // Only pops it off if the top of the stack contains ctx |
436 | bool XLIFFHandler::popContext(XliffContext ctx) |
437 | { |
438 | if (!m_contextStack.isEmpty() && m_contextStack.top() == ctx) { |
439 | m_contextStack.pop(); |
440 | return true; |
441 | } |
442 | return false; |
443 | } |
444 | |
445 | XLIFFHandler::XliffContext XLIFFHandler::currentContext() const |
446 | { |
447 | if (!m_contextStack.isEmpty()) |
448 | return (XliffContext)m_contextStack.top(); |
449 | return XC_xliff; |
450 | } |
451 | |
452 | // traverses to the top to check all of the parent contexes. |
453 | bool XLIFFHandler::hasContext(XliffContext ctx) const |
454 | { |
455 | for (int i = m_contextStack.size() - 1; i >= 0; --i) { |
456 | if (m_contextStack.at(i) == ctx) |
457 | return true; |
458 | } |
459 | return false; |
460 | } |
461 | |
462 | bool XLIFFHandler::startElement(QStringView namespaceURI, QStringView localName, |
463 | QStringView qName, const QXmlStreamAttributes &atts) |
464 | { |
465 | Q_UNUSED(qName); |
466 | if (namespaceURI == m_URITT) |
467 | goto bail; |
468 | if (namespaceURI != m_URI && namespaceURI != m_URI12) { |
469 | return fatalError(line: reader.lineNumber(), column: reader.columnNumber(), |
470 | message: QLatin1String("Unknown namespace in the XLIFF file")); |
471 | } |
472 | if (localName == QLatin1String("xliff")) { |
473 | // make sure that the stack is not empty during parsing |
474 | pushContext(ctx: XC_xliff); |
475 | } else if (localName == QLatin1String("file")) { |
476 | m_fileName = atts.value(qualifiedName: QLatin1String("original")).toString(); |
477 | m_language = atts.value(qualifiedName: QLatin1String("target-language")).toString(); |
478 | m_language.replace(before: QLatin1Char('-'), after: QLatin1Char('_')); |
479 | m_sourceLanguage = atts.value(qualifiedName: QLatin1String("source-language")).toString(); |
480 | m_sourceLanguage.replace(before: QLatin1Char('-'), after: QLatin1Char('_')); |
481 | if (m_sourceLanguage == QLatin1String("en")) |
482 | m_sourceLanguage.clear(); |
483 | } else if (localName == QLatin1String("group")) { |
484 | if (atts.value(qualifiedName: QLatin1String("restype")) == QLatin1String(restypeContext)) { |
485 | m_context = atts.value(qualifiedName: QLatin1String("resname")).toString(); |
486 | pushContext(ctx: XC_restype_context); |
487 | } else { |
488 | if (atts.value(qualifiedName: QLatin1String("restype")) == QLatin1String(restypePlurals)) { |
489 | pushContext(ctx: XC_restype_plurals); |
490 | m_id = atts.value(qualifiedName: QLatin1String("id")).toString(); |
491 | if (atts.value(qualifiedName: QLatin1String("translate")) == QLatin1String( "no")) |
492 | m_translate = false; |
493 | } else { |
494 | pushContext(ctx: XC_group); |
495 | } |
496 | } |
497 | } else if (localName == QLatin1String("trans-unit")) { |
498 | if (!hasContext(ctx: XC_restype_plurals) || m_sources.isEmpty() /* who knows ... */) |
499 | if (atts.value(qualifiedName: QLatin1String("translate")) == QLatin1String( "no")) |
500 | m_translate = false; |
501 | if (!hasContext(ctx: XC_restype_plurals)) { |
502 | m_id = atts.value(qualifiedName: QLatin1String("id")).toString(); |
503 | if (m_id.startsWith(s: QLatin1String("_msg"))) |
504 | m_id.clear(); |
505 | } |
506 | if (atts.value(qualifiedName: QLatin1String("approved")) != QLatin1String( "yes")) |
507 | m_approved = false; |
508 | pushContext(ctx: XC_trans_unit); |
509 | m_hadAlt = false; |
510 | } else if (localName == QLatin1String("alt-trans")) { |
511 | pushContext(ctx: XC_alt_trans); |
512 | } else if (localName == QLatin1String("source")) { |
513 | m_isPlural = atts.value(qualifiedName: QLatin1String(attribPlural)) == QLatin1String("yes"); |
514 | } else if (localName == QLatin1String("target")) { |
515 | if (atts.value(qualifiedName: QLatin1String("restype")) != QLatin1String(restypeDummy)) |
516 | pushContext(ctx: XC_restype_translation); |
517 | } else if (localName == QLatin1String("mrk")) { |
518 | if (atts.value(qualifiedName: QLatin1String("mtype")) == QLatin1String( "seg")) { |
519 | if (currentContext() == XC_restype_translation) |
520 | pushContext(ctx: XC_mtype_seg_translation); |
521 | } |
522 | } else if (localName == QLatin1String("context-group")) { |
523 | if (atts.value(qualifiedName: QLatin1String("purpose")) == QLatin1String( "location")) |
524 | pushContext(ctx: XC_context_group); |
525 | else |
526 | pushContext(ctx: XC_context_group_any); |
527 | } else if (currentContext() == XC_context_group && localName == QLatin1String("context")) { |
528 | const auto ctxtype = atts.value(qualifiedName: QLatin1String("context-type")); |
529 | if (ctxtype == QLatin1String("linenumber")) |
530 | pushContext(ctx: XC_context_linenumber); |
531 | else if (ctxtype == QLatin1String("sourcefile")) |
532 | pushContext(ctx: XC_context_filename); |
533 | } else if (currentContext() == XC_context_group_any && localName == QLatin1String("context")) { |
534 | const auto ctxtype = atts.value(qualifiedName: QLatin1String("context-type")); |
535 | if (ctxtype == QLatin1String(contextMsgctxt)) |
536 | pushContext(ctx: XC_context_comment); |
537 | else if (ctxtype == QLatin1String(contextOldMsgctxt)) |
538 | pushContext(ctx: XC_context_old_comment); |
539 | } else if (localName == QLatin1String("note")) { |
540 | if (atts.value(qualifiedName: QLatin1String("annotates")) == QLatin1String( "source") && |
541 | atts.value(qualifiedName: QLatin1String("from")) == QLatin1String( "developer")) |
542 | pushContext(ctx: XC_extra_comment); |
543 | else |
544 | pushContext(ctx: XC_translator_comment); |
545 | } else if (localName == QLatin1String("ph")) { |
546 | QString ctype = atts.value(qualifiedName: QLatin1String("ctype")).toString(); |
547 | if (ctype.startsWith(s: QLatin1String("x-ch-"))) |
548 | m_ctype = ctype.mid(position: 5); |
549 | pushContext(ctx: XC_ph); |
550 | } |
551 | bail: |
552 | if (currentContext() != XC_ph && currentContext() != XC_mtype_seg_translation) |
553 | accum.clear(); |
554 | return true; |
555 | } |
556 | |
557 | bool XLIFFHandler::endElement(QStringView namespaceURI, QStringView localName, |
558 | QStringView qName) |
559 | { |
560 | Q_UNUSED(qName); |
561 | if (namespaceURI == m_URITT) { |
562 | if (hasContext(ctx: XC_trans_unit) || hasContext(ctx: XC_restype_plurals)) |
563 | m_extra[localName.toString()] = accum; |
564 | else |
565 | m_translator.setExtra(ba: localName.toString(), var: accum); |
566 | return true; |
567 | } |
568 | if (namespaceURI != m_URI && namespaceURI != m_URI12) { |
569 | return fatalError(line: reader.lineNumber(), column: reader.columnNumber(), |
570 | message: QLatin1String("Unknown namespace in the XLIFF file")); |
571 | } |
572 | //qDebug() << "URI:" << namespaceURI << "QNAME:" << qName; |
573 | if (localName == QLatin1String("xliff")) { |
574 | popContext(ctx: XC_xliff); |
575 | } else if (localName == QLatin1String("source")) { |
576 | if (hasContext(ctx: XC_alt_trans)) { |
577 | if (m_isPlural && m_oldSources.isEmpty()) |
578 | m_oldSources.append(t: QString()); |
579 | m_oldSources.append(t: accum); |
580 | m_hadAlt = true; |
581 | } else { |
582 | m_sources.append(t: accum); |
583 | } |
584 | } else if (localName == QLatin1String("target")) { |
585 | if (popContext(ctx: XC_restype_translation)) { |
586 | accum.replace(before: QChar(Translator::TextVariantSeparator), |
587 | after: QChar(Translator::BinaryVariantSeparator)); |
588 | m_translations.append(t: accum); |
589 | } |
590 | } else if (localName == QLatin1String("mrk")) { |
591 | popContext(ctx: XC_mtype_seg_translation); |
592 | } else if (localName == QLatin1String("context-group")) { |
593 | if (popContext(ctx: XC_context_group)) { |
594 | m_refs.append(t: TranslatorMessage::Reference( |
595 | m_extraFileName.isEmpty() ? m_fileName : m_extraFileName, m_lineNumber)); |
596 | m_extraFileName.clear(); |
597 | m_lineNumber = -1; |
598 | } else { |
599 | popContext(ctx: XC_context_group_any); |
600 | } |
601 | } else if (localName == QLatin1String("context")) { |
602 | if (popContext(ctx: XC_context_linenumber)) { |
603 | bool ok; |
604 | m_lineNumber = accum.trimmed().toInt(ok: &ok); |
605 | if (!ok) |
606 | m_lineNumber = -1; |
607 | } else if (popContext(ctx: XC_context_filename)) { |
608 | m_extraFileName = accum; |
609 | } else if (popContext(ctx: XC_context_comment)) { |
610 | m_comment = accum; |
611 | } else if (popContext(ctx: XC_context_old_comment)) { |
612 | m_oldComment = accum; |
613 | } |
614 | } else if (localName == QLatin1String("note")) { |
615 | if (popContext(ctx: XC_extra_comment)) |
616 | m_extraComment = accum; |
617 | else if (popContext(ctx: XC_translator_comment)) |
618 | m_translatorComment = accum; |
619 | } else if (localName == QLatin1String("ph")) { |
620 | m_ctype.clear(); |
621 | popContext(ctx: XC_ph); |
622 | } else if (localName == QLatin1String("trans-unit")) { |
623 | popContext(ctx: XC_trans_unit); |
624 | if (!m_hadAlt) |
625 | m_oldSources.append(t: QString()); |
626 | if (!hasContext(ctx: XC_restype_plurals)) { |
627 | if (!finalizeMessage(isPlural: false)) { |
628 | return fatalError(line: reader.lineNumber(), column: reader.columnNumber(), |
629 | message: QLatin1String("Element processing failed")); |
630 | } |
631 | } |
632 | } else if (localName == QLatin1String("alt-trans")) { |
633 | popContext(ctx: XC_alt_trans); |
634 | } else if (localName == QLatin1String("group")) { |
635 | if (popContext(ctx: XC_restype_plurals)) { |
636 | if (!finalizeMessage(isPlural: true)) { |
637 | return fatalError(line: reader.lineNumber(), column: reader.columnNumber(), |
638 | message: QLatin1String("Element processing failed")); |
639 | } |
640 | } else if (popContext(ctx: XC_restype_context)) { |
641 | m_context.clear(); |
642 | } else { |
643 | popContext(ctx: XC_group); |
644 | } |
645 | } |
646 | return true; |
647 | } |
648 | |
649 | bool XLIFFHandler::characters(QStringView ch) |
650 | { |
651 | if (currentContext() == XC_ph) { |
652 | // handle the content of <ph> elements |
653 | for (int i = 0; i < ch.size(); ++i) { |
654 | QChar chr = ch.at(n: i); |
655 | if (accum.endsWith(c: QLatin1Char('\\'))) |
656 | accum[accum.size() - 1] = QLatin1Char(charFromEscape(escape: chr.toLatin1())); |
657 | else |
658 | accum.append(c: chr); |
659 | } |
660 | } else { |
661 | QString t = ch.toString(); |
662 | t.replace(before: QLatin1String("\r"), after: QLatin1String( "")); |
663 | accum.append(s: t); |
664 | } |
665 | return true; |
666 | } |
667 | |
668 | bool XLIFFHandler::endDocument() |
669 | { |
670 | m_translator.setLanguageCode(m_language); |
671 | m_translator.setSourceLanguageCode(m_sourceLanguage); |
672 | return true; |
673 | } |
674 | |
675 | bool XLIFFHandler::finalizeMessage(bool isPlural) |
676 | { |
677 | if (m_sources.isEmpty()) { |
678 | m_cd.appendError(error: QLatin1String("XLIFF syntax error: Message without source string.")); |
679 | return false; |
680 | } |
681 | if (!m_translate && m_refs.size() == 1 |
682 | && m_refs.at(i: 0).fileName() == QLatin1String(MAGIC_OBSOLETE_REFERENCE)) |
683 | m_refs.clear(); |
684 | TranslatorMessage::Type type |
685 | = m_translate ? (m_approved ? TranslatorMessage::Finished : TranslatorMessage::Unfinished) |
686 | : (m_approved ? TranslatorMessage::Vanished : TranslatorMessage::Obsolete); |
687 | TranslatorMessage msg(m_context, m_sources[0], |
688 | m_comment, QString(), QString(), -1, |
689 | m_translations, type, isPlural); |
690 | msg.setId(m_id); |
691 | msg.setReferences(m_refs); |
692 | msg.setOldComment(m_oldComment); |
693 | msg.setExtraComment(m_extraComment); |
694 | msg.setTranslatorComment(m_translatorComment); |
695 | msg.setFileName(m_fileName); |
696 | if (m_sources.size() > 1 && m_sources[1] != m_sources[0]) |
697 | m_extra.insert(key: QLatin1String("po-msgid_plural"), value: m_sources[1]); |
698 | if (!m_oldSources.isEmpty()) { |
699 | if (!m_oldSources[0].isEmpty()) |
700 | msg.setOldSourceText(m_oldSources[0]); |
701 | if (m_oldSources.size() > 1 && m_oldSources[1] != m_oldSources[0]) |
702 | m_extra.insert(key: QLatin1String("po-old_msgid_plural"), value: m_oldSources[1]); |
703 | } |
704 | msg.setExtras(m_extra); |
705 | m_translator.append(msg); |
706 | |
707 | m_id.clear(); |
708 | m_sources.clear(); |
709 | m_oldSources.clear(); |
710 | m_translations.clear(); |
711 | m_comment.clear(); |
712 | m_oldComment.clear(); |
713 | m_extraComment.clear(); |
714 | m_translatorComment.clear(); |
715 | m_extra.clear(); |
716 | m_refs.clear(); |
717 | m_translate = true; |
718 | m_approved = true; |
719 | return true; |
720 | } |
721 | |
722 | bool XLIFFHandler::fatalError(qint64 line, qint64 column, const QString &message) |
723 | { |
724 | QString msg = QString::asprintf(format: "XML error: Parse error at line %d, column %d (%s).\n", |
725 | static_cast<int>(line), static_cast<int>(column), |
726 | message.toLatin1().data()); |
727 | m_cd.appendError(error: msg); |
728 | return false; |
729 | } |
730 | |
731 | bool loadXLIFF(Translator &translator, QIODevice &dev, ConversionData &cd) |
732 | { |
733 | QXmlStreamReader reader(&dev); |
734 | XLIFFHandler hand(translator, cd, reader); |
735 | return hand.parse(); |
736 | } |
737 | |
738 | bool saveXLIFF(const Translator &translator, QIODevice &dev, ConversionData &cd) |
739 | { |
740 | bool ok = true; |
741 | int indent = 0; |
742 | |
743 | QTextStream ts(&dev); |
744 | |
745 | QStringList dtgs = cd.dropTags(); |
746 | dtgs << QLatin1String("po-(old_)?msgid_plural"); |
747 | QRegularExpression drops(QRegularExpression::anchoredPattern(expression: dtgs.join(sep: QLatin1Char('|')))); |
748 | |
749 | QHash<QString, QHash<QString, QList<TranslatorMessage> > > messageOrder; |
750 | QHash<QString, QList<QString> > contextOrder; |
751 | QList<QString> fileOrder; |
752 | for (const TranslatorMessage &msg : translator.messages()) { |
753 | QString fn = msg.fileName(); |
754 | if (fn.isEmpty() && msg.type() == TranslatorMessage::Obsolete) |
755 | fn = QLatin1String(MAGIC_OBSOLETE_REFERENCE); |
756 | QHash<QString, QList<TranslatorMessage> > &file = messageOrder[fn]; |
757 | if (file.isEmpty()) |
758 | fileOrder.append(t: fn); |
759 | QList<TranslatorMessage> &context = file[msg.context()]; |
760 | if (context.isEmpty()) |
761 | contextOrder[fn].append(t: msg.context()); |
762 | context.append(t: msg); |
763 | } |
764 | |
765 | ts.setFieldAlignment(QTextStream::AlignRight); |
766 | ts << "<?xml version=\"1.0\""; |
767 | ts << " encoding=\"utf-8\"?>\n"; |
768 | ts << "<xliff version=\"1.2\" xmlns=\""<< XLIFF12namespaceURI |
769 | << "\" xmlns:trolltech=\""<< TrollTsNamespaceURI << "\">\n"; |
770 | ++indent; |
771 | writeExtras(ts, indent, extras: translator.extras(), drops); |
772 | QString sourceLanguageCode = translator.sourceLanguageCode(); |
773 | if (sourceLanguageCode.isEmpty() || sourceLanguageCode == QLatin1String("C")) |
774 | sourceLanguageCode = QLatin1String("en"); |
775 | else |
776 | sourceLanguageCode.replace(before: QLatin1Char('_'), after: QLatin1Char('-')); |
777 | QString languageCode = translator.languageCode(); |
778 | languageCode.replace(before: QLatin1Char('_'), after: QLatin1Char('-')); |
779 | for (const QString &fn : std::as_const(t&: fileOrder)) { |
780 | writeIndent(ts, indent); |
781 | ts << "<file original=\""<< fn << "\"" |
782 | << " datatype=\""<< dataType(m: messageOrder[fn].cbegin()->first()) << "\"" |
783 | << " source-language=\""<< sourceLanguageCode.toLatin1() << "\"" |
784 | << " target-language=\""<< languageCode.toLatin1() << "\"" |
785 | << "><body>\n"; |
786 | ++indent; |
787 | |
788 | for (const QString &ctx : std::as_const(t&: contextOrder[fn])) { |
789 | if (!ctx.isEmpty()) { |
790 | writeIndent(ts, indent); |
791 | ts << "<group restype=\""<< restypeContext << "\"" |
792 | << " resname=\""<< xlProtect(str: ctx) << "\">\n"; |
793 | ++indent; |
794 | } |
795 | |
796 | for (const TranslatorMessage &msg : std::as_const(t&: messageOrder[fn][ctx])) |
797 | writeMessage(ts, msg, drops, indent); |
798 | |
799 | if (!ctx.isEmpty()) { |
800 | --indent; |
801 | writeIndent(ts, indent); |
802 | ts << "</group>\n"; |
803 | } |
804 | } |
805 | |
806 | --indent; |
807 | writeIndent(ts, indent); |
808 | ts << "</body></file>\n"; |
809 | } |
810 | --indent; |
811 | writeIndent(ts, indent); |
812 | ts << "</xliff>\n"; |
813 | |
814 | return ok; |
815 | } |
816 | |
817 | int initXLIFF() |
818 | { |
819 | Translator::FileFormat format; |
820 | format.extension = QLatin1String("xlf"); |
821 | format.untranslatedDescription = QT_TRANSLATE_NOOP("FMT", "XLIFF localization files"); |
822 | format.fileType = Translator::FileFormat::TranslationSource; |
823 | format.priority = 1; |
824 | format.loader = &loadXLIFF; |
825 | format.saver = &saveXLIFF; |
826 | Translator::registerFileFormat(format); |
827 | return 1; |
828 | } |
829 | |
830 | Q_CONSTRUCTOR_FUNCTION(initXLIFF) |
831 | |
832 | QT_END_NAMESPACE |
833 |
Definitions
- restypeContext
- restypePlurals
- restypeDummy
- dataTypeUIFile
- contextMsgctxt
- contextOldMsgctxt
- attribPlural
- XLIFF11namespaceURI
- XLIFF12namespaceURI
- TrollTsNamespaceURI
- dataType
- writeIndent
- CharMnemonic
- charCodeMnemonics
- charFromEscape
- xlNumericEntity
- xlProtect
- writeExtras
- writeLineNumber
- writeComment
- writeTransUnits
- writeMessage
- XLIFFHandler
- ~XLIFFHandler
- XliffContext
- XLIFFHandler
- pushContext
- popContext
- currentContext
- hasContext
- startElement
- endElement
- characters
- endDocument
- finalizeMessage
- fatalError
- loadXLIFF
- saveXLIFF
Learn to use CMake with our Intro Training
Find out more