src/Language.cpp

   1 /**
   2  * \file Language.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author Lars Gullik Bjønnes
   7  * \author Jean-Marc Lasgouttes
   8  * \author Jürgen Spitzmüller
   9  * \author Dekel Tsur
  10  *
  11  * Full author contact details are available in file CREDITS.
  12  */
  13
  14 #include <config.h>
  15
  16 #include "Language.h"
  17
  18 #include "Encoding.h"
  19 #include "LaTeXFonts.h"
  20 #include "Lexer.h"
  21 #include "LyXRC.h"
  22
  23 #include "support/debug.h"
  24 #include "support/FileName.h"
  25 #include "support/filetools.h"
  26 #include "support/lassert.h"
  27 #include "support/lstrings.h"
  28 #include "support/qstring_helpers.h"
  29 #include "support/Messages.h"
  30
  31 #include <QLocale>
  32 #include <QString>
  33
  34 using namespace std;
  35 using namespace lyx::support;
  36
  37 namespace lyx {
  38
  39 Languages languages;
  40 Language const * ignore_language = 0;
  41 Language const * default_language = 0;
  42 Language const * latex_language = 0;
  43 Language const * reset_language = 0;
  44
  45
  46 bool Language::isPolyglossiaExclusive() const
  47 {
  48         return babel().empty() && !polyglossia().empty() && required().empty();
  49 }
  50
  51
  52 bool Language::isBabelExclusive() const
  53 {
  54         return !babel().empty() && polyglossia().empty() && required().empty();
  55 }
  56
  57
  58 docstring const Language::translateLayout(string const & m) const
  59 {
  60         if (m.empty())
  61                 return docstring();
  62
  63         if (!isAscii(m)) {
  64                 lyxerr << "Warning: not translating `" << m
  65                        << "' because it is not pure ASCII.\n";
  66                 return from_utf8(m);
  67         }
  68
  69         TranslationMap::const_iterator it = layoutTranslations_.find(m);
  70         if (it != layoutTranslations_.end())
  71                 return it->second;
  72
  73         docstring t = from_ascii(m);
  74         cleanTranslation(t);
  75         return t;
  76 }
  77
  78
  79 string Language::fontenc(BufferParams const & params) const
  80 {
  81         // Don't use LaTeX fonts, so just return the language's preferred
  82         // (although this is not used with nonTeXFonts anyway).
  83         if (params.useNonTeXFonts)
  84                 return fontenc_.front() == "ASCII" ? "T1" : fontenc_.front();
  85
  86         // Determine optimal font encoding
  87         // We check whether the used rm font supports an encoding our language supports
  88         LaTeXFont const & lf =
  89                 theLaTeXFonts().getLaTeXFont(from_ascii(params.fontsRoman()));
  90         vector<string> const lfe = lf.fontencs();
  91         for (auto & fe : fontenc_) {
  92                 // ASCII means: support all T* encodings plus OT1
  93                 if (fe == "ASCII") {
  94                         for (auto & afe : lfe) {
  95                                 if (afe == "OT1" || prefixIs(afe, "T"))
  96                                         // we found a suitable one; return that.
  97                                         return afe;
  98                         }
  99                 }
 100                 // For other encodings, just check whether the font supports it
 101                 if (lf.hasFontenc(fe))
 102                         return fe;
 103         }
 104         // We did not find a suitable one; just take the first in the list,
 105         // the priorized one (which is "T1" for ASCII).
 106         return fontenc_.front() == "ASCII" ? "T1" : fontenc_.front();
 107 }
 108
 109
 110 string Language::dateFormat(size_t i) const
 111 {
 112         if (i > dateformats_.size())
 113                 return string();
 114         return dateformats_.at(i);
 115 }
 116
 117
 118 docstring Language::decimalSeparator() const
 119 {
 120         if (lyxrc.default_decimal_sep == "locale") {
 121                 QLocale loc = QLocale(toqstr(code()));
 122                 return qstring_to_ucs4(QString(loc.decimalPoint()));
 123         }
 124         return from_utf8(lyxrc.default_decimal_sep);
 125 }
 126
 127
 128 bool Language::readLanguage(Lexer & lex)
 129 {
 130         enum LanguageTags {
 131                 LA_BABELNAME = 1,
 132                 LA_DATEFORMATS,
 133                 LA_ENCODING,
 134                 LA_END,
 135                 LA_FONTENC,
 136                 LA_GUINAME,
 137                 LA_HAS_GUI_SUPPORT,
 138                 LA_INTERNAL_ENC,
 139                 LA_LANG_CODE,
 140                 LA_LANG_VARIETY,
 141                 LA_POLYGLOSSIANAME,
 142                 LA_POLYGLOSSIAOPTS,
 143                 LA_XINDYNAME,
 144                 LA_POSTBABELPREAMBLE,
 145                 LA_PREBABELPREAMBLE,
 146                 LA_PROVIDES,
 147                 LA_REQUIRES,
 148                 LA_QUOTESTYLE,
 149                 LA_RTL,
 150                 LA_WORDWRAP,
 151                 LA_ACTIVECHARS
 152         };
 153
 154         // Keep these sorted alphabetically!
 155         LexerKeyword languageTags[] = {
 156                 { "activechars",          LA_ACTIVECHARS },
 157                 { "babelname",            LA_BABELNAME },
 158                 { "dateformats",          LA_DATEFORMATS },
 159                 { "encoding",             LA_ENCODING },
 160                 { "end",                  LA_END },
 161                 { "fontencoding",         LA_FONTENC },
 162                 { "guiname",              LA_GUINAME },
 163                 { "hasguisupport",        LA_HAS_GUI_SUPPORT },
 164                 { "internalencoding",     LA_INTERNAL_ENC },
 165                 { "langcode",             LA_LANG_CODE },
 166                 { "langvariety",          LA_LANG_VARIETY },
 167                 { "polyglossianame",      LA_POLYGLOSSIANAME },
 168                 { "polyglossiaopts",      LA_POLYGLOSSIAOPTS },
 169                 { "postbabelpreamble",    LA_POSTBABELPREAMBLE },
 170                 { "prebabelpreamble",     LA_PREBABELPREAMBLE },
 171                 { "provides",             LA_PROVIDES },
 172                 { "quotestyle",           LA_QUOTESTYLE },
 173                 { "requires",             LA_REQUIRES },
 174                 { "rtl",                  LA_RTL },
 175                 { "wordwrap",             LA_WORDWRAP },
 176                 { "xindyname",            LA_XINDYNAME }
 177         };
 178
 179         bool error = false;
 180         bool finished = false;
 181         lex.pushTable(languageTags);
 182         // parse style section
 183         while (!finished && lex.isOK() && !error) {
 184                 int le = lex.lex();
 185                 // See comment in LyXRC.cpp.
 186                 switch (le) {
 187                 case Lexer::LEX_FEOF:
 188                         continue;
 189
 190                 case Lexer::LEX_UNDEF: // parse error
 191                         lex.printError("Unknown language tag `$$Token'");
 192                         error = true;
 193                         continue;
 194
 195                 default:
 196                         break;
 197                 }
 198                 switch (static_cast<LanguageTags>(le)) {
 199                 case LA_END: // end of structure
 200                         finished = true;
 201                         break;
 202                 case LA_BABELNAME:
 203                         lex >> babel_;
 204                         break;
 205                 case LA_POLYGLOSSIANAME:
 206                         lex >> polyglossia_name_;
 207                         break;
 208                 case LA_POLYGLOSSIAOPTS:
 209                         lex >> polyglossia_opts_;
 210                         break;
 211                 case LA_XINDYNAME:
 212                         lex >> xindy_;
 213                         break;
 214                 case LA_QUOTESTYLE:
 215                         lex >> quote_style_;
 216                         break;
 217                 case LA_ACTIVECHARS:
 218                         lex >> active_chars_;
 219                         break;
 220                 case LA_ENCODING:
 221                         lex >> encodingStr_;
 222                         break;
 223                 case LA_FONTENC: {
 224                         lex.eatLine();
 225                         vector<string> const fe =
 226                                 getVectorFromString(lex.getString(true), "|");
 227                         fontenc_.insert(fontenc_.end(), fe.begin(), fe.end());
 228                         break;
 229                 }
 230                 case LA_DATEFORMATS: {
 231                         lex.eatLine();
 232                         vector<string> const df =
 233                                 getVectorFromString(trim(lex.getString(true), "\""), "|");
 234                         dateformats_.insert(dateformats_.end(), df.begin(), df.end());
 235                         break;
 236                 }
 237                 case LA_GUINAME:
 238                         lex >> display_;
 239                         break;
 240                 case LA_HAS_GUI_SUPPORT:
 241                         lex >> has_gui_support_;
 242                         break;
 243                 case LA_INTERNAL_ENC:
 244                         lex >> internal_enc_;
 245                         break;
 246                 case LA_LANG_CODE:
 247                         lex >> code_;
 248                         break;
 249                 case LA_LANG_VARIETY:
 250                         lex >> variety_;
 251                         break;
 252                 case LA_POSTBABELPREAMBLE:
 253                         babel_postsettings_ =
 254                                 lex.getLongString(from_ascii("EndPostBabelPreamble"));
 255                         break;
 256                 case LA_PREBABELPREAMBLE:
 257                         babel_presettings_ =
 258                                 lex.getLongString(from_ascii("EndPreBabelPreamble"));
 259                         break;
 260                 case LA_REQUIRES:
 261                         lex >> required_;
 262                         break;
 263                 case LA_PROVIDES:
 264                         lex >> provides_;
 265                         break;
 266                 case LA_RTL:
 267                         lex >> rightToLeft_;
 268                         break;
 269                 case LA_WORDWRAP:
 270                         lex >> word_wrap_;
 271                         break;
 272                 }
 273         }
 274         lex.popTable();
 275         return finished && !error;
 276 }
 277
 278
 279 bool Language::read(Lexer & lex)
 280 {
 281         encoding_ = nullptr;
 282         internal_enc_ = false;
 283         rightToLeft_ = false;
 284
 285         if (!lex.next()) {
 286                 lex.printError("No name given for language: `$$Token'.");
 287                 return false;
 288         }
 289
 290         lang_ = lex.getString();
 291         LYXERR(Debug::INFO, "Reading language " << lang_);
 292         if (!readLanguage(lex)) {
 293                 LYXERR0("Error parsing language `" << lang_ << '\'');
 294                 return false;
 295         }
 296
 297         encoding_ = encodings.fromLyXName(encodingStr_);
 298         if (!encoding_ && !encodingStr_.empty()) {
 299                 encoding_ = encodings.fromLyXName("iso8859-1");
 300                 LYXERR0("Unknown encoding " << encodingStr_);
 301         }
 302         if (fontenc_.empty())
 303                 fontenc_.push_back("ASCII");
 304         if (dateformats_.empty()) {
 305                 dateformats_.push_back("MMMM dd, yyyy");
 306                 dateformats_.push_back("MMM dd, yyyy");
 307                 dateformats_.push_back("M/d/yyyy");
 308         }
 309         return true;
 310 }
 311
 312
 313 void Language::readLayoutTranslations(Language::TranslationMap const & trans, bool replace)
 314 {
 315         for (auto const & t : trans) {
 316                 if (replace
 317                     || layoutTranslations_.find(t.first) == layoutTranslations_.end())
 318                         layoutTranslations_[t.first] = t.second;
 319         }
 320 }
 321
 322
 323 void Languages::read(FileName const & filename)
 324 {
 325         Lexer lex;
 326         lex.setFile(filename);
 327         lex.setContext("Languages::read");
 328         while (lex.isOK()) {
 329                 int le = lex.lex();
 330                 switch (le) {
 331                 case Lexer::LEX_FEOF:
 332                         continue;
 333
 334                 default:
 335                         break;
 336                 }
 337                 if (lex.getString() != "Language") {
 338                         lex.printError("Unknown Language tag `$$Token'");
 339                         continue;
 340                 }
 341                 Language l;
 342                 l.read(lex);
 343                 if (!lex)
 344                         break;
 345                 if (l.lang() == "latex") {
 346                         // Check if latex language was not already defined.
 347                         LASSERT(latex_language == 0, continue);
 348                         static const Language latex_lang = l;
 349                         latex_language = &latex_lang;
 350                 } else if (l.lang() == "ignore") {
 351                         // Check if ignore language was not already defined.
 352                         LASSERT(ignore_language == 0, continue);
 353                         static const Language ignore_lang = l;
 354                         ignore_language = &ignore_lang;
 355                 } else
 356                         languagelist_[l.lang()] = l;
 357         }
 358
 359         default_language = getLanguage("english");
 360         if (!default_language) {
 361                 LYXERR0("Default language \"english\" not found!");
 362                 default_language = &(*languagelist_.begin()).second;
 363                 LYXERR0("Using \"" << default_language->lang() << "\" instead!");
 364         }
 365
 366         // Read layout translations
 367         FileName const path = libFileSearch(string(), "layouttranslations");
 368         readLayoutTranslations(path);
 369 }
 370
 371
 372 namespace {
 373
 374 bool readTranslations(Lexer & lex, Language::TranslationMap & trans)
 375 {
 376         while (lex.isOK()) {
 377                 if (lex.checkFor("End"))
 378                         break;
 379                 if (!lex.next(true))
 380                         return false;
 381                 string const key = lex.getString();
 382                 if (!lex.next(true))
 383                         return false;
 384                 docstring const val = lex.getDocString();
 385                 trans[key] = val;
 386         }
 387         return true;
 388 }
 389
 390
 391 enum Match {
 392         NoMatch,
 393         ApproximateMatch,
 394         VeryApproximateMatch,
 395         ExactMatch
 396 };
 397
 398
 399 Match match(string const & code, Language const & lang)
 400 {
 401         // we need to mimic gettext: code can be a two-letter code, which
 402         // should match all variants, e.g. "de" should match "de_DE",
 403         // "de_AT" etc.
 404         // special case for chinese:
 405         // simplified  => code == "zh_CN", langcode == "zh_CN"
 406         // traditional => code == "zh_TW", langcode == "zh_CN"
 407         string const variety = lang.variety();
 408         string const langcode = variety.empty() ?
 409                                 lang.code() : lang.code() + '_' + variety;
 410         string const name = lang.lang();
 411         if ((code == langcode && name != "chinese-traditional")
 412                 || (code == "zh_TW"  && name == "chinese-traditional"))
 413                 return ExactMatch;
 414         if ((code.size() == 2) && (langcode.size() > 2)
 415                 && (code + '_' == langcode.substr(0, 3)))
 416                 return ApproximateMatch;
 417         if (code.substr(0,2) == langcode.substr(0,2))
 418                 return VeryApproximateMatch;
 419         return NoMatch;
 420 }
 421
 422 } // namespace
 423
 424
 425
 426 Language const * Languages::getFromCode(string const & code) const
 427 {
 428         // 1/ exact match with any known language
 429         for (auto const & l : languagelist_) {
 430                 if (match(code, l.second) == ExactMatch)
 431                         return &l.second;
 432         }
 433
 434         // 2/ approximate with any known language
 435         for (auto const & l : languagelist_) {
 436                 if (match(code, l.second) == ApproximateMatch)
 437                         return &l.second;
 438         }
 439         return 0;
 440 }
 441
 442
 443 Language const * Languages::getFromCode(string const & code,
 444                         set<Language const *> const & tryfirst) const
 445 {
 446         // 1/ exact match with tryfirst list
 447         for (auto const * lptr : tryfirst) {
 448                 if (match(code, *lptr) == ExactMatch)
 449                         return lptr;
 450         }
 451
 452         // 2/ approximate match with tryfirst list
 453         for (auto const * lptr : tryfirst) {
 454                 Match const m = match(code, *lptr);
 455                 if (m == ApproximateMatch || m == VeryApproximateMatch)
 456                         return lptr;
 457         }
 458
 459         // 3/ stricter match in all languages
 460         return getFromCode(code);
 461
 462         LYXERR0("Unknown language `" << code << "'");
 463         return 0;
 464 }
 465
 466
 467 void Languages::readLayoutTranslations(support::FileName const & filename)
 468 {
 469         Lexer lex;
 470         lex.setFile(filename);
 471         lex.setContext("Languages::read");
 472
 473         // 1) read all translations (exact and approximate matches) into trans
 474         std::map<string, Language::TranslationMap> trans;
 475         while (lex.isOK()) {
 476                 if (!lex.checkFor("Translation")) {
 477                         if (lex.isOK())
 478                                 lex.printError("Unknown layout translation tag `$$Token'");
 479                         break;
 480                 }
 481                 if (!lex.next(true))
 482                         break;
 483                 string const code = lex.getString();
 484                 bool found = getFromCode(code);
 485                 if (!found) {
 486                         lex.printError("Unknown language `" + code + "'");
 487                         break;
 488                 }
 489                 if (!readTranslations(lex, trans[code])) {
 490                         lex.printError("Could not read layout translations for language `"
 491                                 + code + "'");
 492                         break;
 493                 }
 494         }
 495
 496         // 2) merge all translations into the languages
 497         // exact translations overwrite approximate ones
 498         for (auto & tr : trans) {
 499                 for (auto & lang : languagelist_) {
 500                         Match const m = match(tr.first, lang.second);
 501                         if (m == NoMatch)
 502                                 continue;
 503                         lang.second.readLayoutTranslations(tr.second, m == ExactMatch);
 504                 }
 505         }
 506
 507 }
 508
 509
 510 Language const * Languages::getLanguage(string const & language) const
 511 {
 512         if (language == "reset")
 513                 return reset_language;
 514         if (language == "ignore")
 515                 return ignore_language;
 516         const_iterator it = languagelist_.find(language);
 517         return it == languagelist_.end() ? reset_language : &it->second;
 518 }
 519
 520
 521 } // namespace lyx