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() && requires().empty();
  49 }
  50
  51
  52 bool Language::isBabelExclusive() const
  53 {
  54         return !babel().empty() && polyglossia().empty() && requires().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_POSTBABELPREAMBLE,
 144                 LA_PREBABELPREAMBLE,
 145                 LA_PROVIDES,
 146                 LA_REQUIRES,
 147                 LA_QUOTESTYLE,
 148                 LA_RTL,
 149                 LA_WORDWRAP
 150         };
 151
 152         // Keep these sorted alphabetically!
 153         LexerKeyword languageTags[] = {
 154                 { "babelname",            LA_BABELNAME },
 155                 { "dateformats",          LA_DATEFORMATS },
 156                 { "encoding",             LA_ENCODING },
 157                 { "end",                  LA_END },
 158                 { "fontencoding",         LA_FONTENC },
 159                 { "guiname",              LA_GUINAME },
 160                 { "hasguisupport",        LA_HAS_GUI_SUPPORT },
 161                 { "internalencoding",     LA_INTERNAL_ENC },
 162                 { "langcode",             LA_LANG_CODE },
 163                 { "langvariety",          LA_LANG_VARIETY },
 164                 { "polyglossianame",      LA_POLYGLOSSIANAME },
 165                 { "polyglossiaopts",      LA_POLYGLOSSIAOPTS },
 166                 { "postbabelpreamble",    LA_POSTBABELPREAMBLE },
 167                 { "prebabelpreamble",     LA_PREBABELPREAMBLE },
 168                 { "provides",             LA_PROVIDES },
 169                 { "quotestyle",           LA_QUOTESTYLE },
 170                 { "requires",             LA_REQUIRES },
 171                 { "rtl",                  LA_RTL },
 172                 {"wordwrap",              LA_WORDWRAP }
 173         };
 174
 175         bool error = false;
 176         bool finished = false;
 177         lex.pushTable(languageTags);
 178         // parse style section
 179         while (!finished && lex.isOK() && !error) {
 180                 int le = lex.lex();
 181                 // See comment in LyXRC.cpp.
 182                 switch (le) {
 183                 case Lexer::LEX_FEOF:
 184                         continue;
 185
 186                 case Lexer::LEX_UNDEF: // parse error
 187                         lex.printError("Unknown language tag `$$Token'");
 188                         error = true;
 189                         continue;
 190
 191                 default:
 192                         break;
 193                 }
 194                 switch (static_cast<LanguageTags>(le)) {
 195                 case LA_END: // end of structure
 196                         finished = true;
 197                         break;
 198                 case LA_BABELNAME:
 199                         lex >> babel_;
 200                         break;
 201                 case LA_POLYGLOSSIANAME:
 202                         lex >> polyglossia_name_;
 203                         break;
 204                 case LA_POLYGLOSSIAOPTS:
 205                         lex >> polyglossia_opts_;
 206                         break;
 207                 case LA_QUOTESTYLE:
 208                         lex >> quote_style_;
 209                         break;
 210                 case LA_ENCODING:
 211                         lex >> encodingStr_;
 212                         break;
 213                 case LA_FONTENC: {
 214                         lex.eatLine();
 215                         vector<string> const fe =
 216                                 getVectorFromString(lex.getString(true), "|");
 217                         fontenc_.insert(fontenc_.end(), fe.begin(), fe.end());
 218                         break;
 219                 }
 220                 case LA_DATEFORMATS: {
 221                         lex.eatLine();
 222                         vector<string> const df =
 223                                 getVectorFromString(trim(lex.getString(true), "\""), "|");
 224                         dateformats_.insert(dateformats_.end(), df.begin(), df.end());
 225                         break;
 226                 }
 227                 case LA_GUINAME:
 228                         lex >> display_;
 229                         break;
 230                 case LA_HAS_GUI_SUPPORT:
 231                         lex >> has_gui_support_;
 232                         break;
 233                 case LA_INTERNAL_ENC:
 234                         lex >> internal_enc_;
 235                         break;
 236                 case LA_LANG_CODE:
 237                         lex >> code_;
 238                         break;
 239                 case LA_LANG_VARIETY:
 240                         lex >> variety_;
 241                         break;
 242                 case LA_POSTBABELPREAMBLE:
 243                         babel_postsettings_ =
 244                                 lex.getLongString(from_ascii("EndPostBabelPreamble"));
 245                         break;
 246                 case LA_PREBABELPREAMBLE:
 247                         babel_presettings_ =
 248                                 lex.getLongString(from_ascii("EndPreBabelPreamble"));
 249                         break;
 250                 case LA_REQUIRES:
 251                         lex >> requires_;
 252                         break;
 253                 case LA_PROVIDES:
 254                         lex >> provides_;
 255                         break;
 256                 case LA_RTL:
 257                         lex >> rightToLeft_;
 258                         break;
 259                 case LA_WORDWRAP:
 260                         lex >> word_wrap_;
 261                         break;
 262                 }
 263         }
 264         lex.popTable();
 265         return finished && !error;
 266 }
 267
 268
 269 bool Language::read(Lexer & lex)
 270 {
 271         encoding_ = 0;
 272         internal_enc_ = 0;
 273         rightToLeft_ = 0;
 274
 275         if (!lex.next()) {
 276                 lex.printError("No name given for language: `$$Token'.");
 277                 return false;
 278         }
 279
 280         lang_ = lex.getString();
 281         LYXERR(Debug::INFO, "Reading language " << lang_);
 282         if (!readLanguage(lex)) {
 283                 LYXERR0("Error parsing language `" << lang_ << '\'');
 284                 return false;
 285         }
 286
 287         encoding_ = encodings.fromLyXName(encodingStr_);
 288         if (!encoding_ && !encodingStr_.empty()) {
 289                 encoding_ = encodings.fromLyXName("iso8859-1");
 290                 LYXERR0("Unknown encoding " << encodingStr_);
 291         }
 292         if (fontenc_.empty())
 293                 fontenc_.push_back("ASCII");
 294         if (dateformats_.empty()) {
 295                 dateformats_.push_back("MMMM dd, yyyy");
 296                 dateformats_.push_back("MMM dd, yyyy");
 297                 dateformats_.push_back("M/d/yyyy");
 298         }
 299         return true;
 300 }
 301
 302
 303 void Language::readLayoutTranslations(Language::TranslationMap const & trans, bool replace)
 304 {
 305         for (auto const & t : trans) {
 306                 if (replace
 307                     || layoutTranslations_.find(t.first) == layoutTranslations_.end())
 308                         layoutTranslations_[t.first] = t.second;
 309         }
 310 }
 311
 312
 313 void Languages::read(FileName const & filename)
 314 {
 315         Lexer lex;
 316         lex.setFile(filename);
 317         lex.setContext("Languages::read");
 318         while (lex.isOK()) {
 319                 int le = lex.lex();
 320                 switch (le) {
 321                 case Lexer::LEX_FEOF:
 322                         continue;
 323
 324                 default:
 325                         break;
 326                 }
 327                 if (lex.getString() != "Language") {
 328                         lex.printError("Unknown Language tag `$$Token'");
 329                         continue;
 330                 }
 331                 Language l;
 332                 l.read(lex);
 333                 if (!lex)
 334                         break;
 335                 if (l.lang() == "latex") {
 336                         // Check if latex language was not already defined.
 337                         LASSERT(latex_language == 0, continue);
 338                         static const Language latex_lang = l;
 339                         latex_language = &latex_lang;
 340                 } else if (l.lang() == "ignore") {
 341                         // Check if ignore language was not already defined.
 342                         LASSERT(ignore_language == 0, continue);
 343                         static const Language ignore_lang = l;
 344                         ignore_language = &ignore_lang;
 345                 } else
 346                         languagelist_[l.lang()] = l;
 347         }
 348
 349         default_language = getLanguage("english");
 350         if (!default_language) {
 351                 LYXERR0("Default language \"english\" not found!");
 352                 default_language = &(*languagelist_.begin()).second;
 353                 LYXERR0("Using \"" << default_language->lang() << "\" instead!");
 354         }
 355
 356         // Read layout translations
 357         FileName const path = libFileSearch(string(), "layouttranslations");
 358         readLayoutTranslations(path);
 359 }
 360
 361
 362 namespace {
 363
 364 bool readTranslations(Lexer & lex, Language::TranslationMap & trans)
 365 {
 366         while (lex.isOK()) {
 367                 if (lex.checkFor("End"))
 368                         break;
 369                 if (!lex.next(true))
 370                         return false;
 371                 string const key = lex.getString();
 372                 if (!lex.next(true))
 373                         return false;
 374                 docstring const val = lex.getDocString();
 375                 trans[key] = val;
 376         }
 377         return true;
 378 }
 379
 380
 381 enum Match {
 382         NoMatch,
 383         ApproximateMatch,
 384         VeryApproximateMatch,
 385         ExactMatch
 386 };
 387
 388
 389 Match match(string const & code, Language const & lang)
 390 {
 391         // we need to mimic gettext: code can be a two-letter code, which
 392         // should match all variants, e.g. "de" should match "de_DE",
 393         // "de_AT" etc.
 394         // special case for chinese:
 395         // simplified  => code == "zh_CN", langcode == "zh_CN"
 396         // traditional => code == "zh_TW", langcode == "zh_CN"
 397         string const variety = lang.variety();
 398         string const langcode = variety.empty() ?
 399                                 lang.code() : lang.code() + '_' + variety;
 400         string const name = lang.lang();
 401         if ((code == langcode && name != "chinese-traditional")
 402                 || (code == "zh_TW"  && name == "chinese-traditional"))
 403                 return ExactMatch;
 404         if ((code.size() == 2) && (langcode.size() > 2)
 405                 && (code + '_' == langcode.substr(0, 3)))
 406                 return ApproximateMatch;
 407         if (code.substr(0,2) == langcode.substr(0,2))
 408                 return VeryApproximateMatch;
 409         return NoMatch;
 410 }
 411
 412 } // namespace
 413
 414
 415
 416 Language const * Languages::getFromCode(string const & code) const
 417 {
 418         // 1/ exact match with any known language
 419         for (auto const & l : languagelist_) {
 420                 if (match(code, l.second) == ExactMatch)
 421                         return &l.second;
 422         }
 423
 424         // 2/ approximate with any known language
 425         for (auto const & l : languagelist_) {
 426                 if (match(code, l.second) == ApproximateMatch)
 427                         return &l.second;
 428         }
 429         return 0;
 430 }
 431
 432
 433 Language const * Languages::getFromCode(string const & code,
 434                         set<Language const *> const & tryfirst) const
 435 {
 436         // 1/ exact match with tryfirst list
 437         for (auto const * lptr : tryfirst) {
 438                 if (match(code, *lptr) == ExactMatch)
 439                         return lptr;
 440         }
 441
 442         // 2/ approximate match with tryfirst list
 443         for (auto const * lptr : tryfirst) {
 444                 Match const m = match(code, *lptr);
 445                 if (m == ApproximateMatch || m == VeryApproximateMatch)
 446                         return lptr;
 447         }
 448
 449         // 3/ stricter match in all languages
 450         return getFromCode(code);
 451
 452         LYXERR0("Unknown language `" << code << "'");
 453         return 0;
 454 }
 455
 456
 457 void Languages::readLayoutTranslations(support::FileName const & filename)
 458 {
 459         Lexer lex;
 460         lex.setFile(filename);
 461         lex.setContext("Languages::read");
 462
 463         // 1) read all translations (exact and approximate matches) into trans
 464         std::map<string, Language::TranslationMap> trans;
 465         while (lex.isOK()) {
 466                 if (!lex.checkFor("Translation")) {
 467                         if (lex.isOK())
 468                                 lex.printError("Unknown layout translation tag `$$Token'");
 469                         break;
 470                 }
 471                 if (!lex.next(true))
 472                         break;
 473                 string const code = lex.getString();
 474                 bool found = getFromCode(code);
 475                 if (!found) {
 476                         lex.printError("Unknown language `" + code + "'");
 477                         break;
 478                 }
 479                 if (!readTranslations(lex, trans[code])) {
 480                         lex.printError("Could not read layout translations for language `"
 481                                 + code + "'");
 482                         break;
 483                 }
 484         }
 485
 486         // 2) merge all translations into the languages
 487         // exact translations overwrite approximate ones
 488         for (auto & tr : trans) {
 489                 for (auto & lang : languagelist_) {
 490                         Match const m = match(tr.first, lang.second);
 491                         if (m == NoMatch)
 492                                 continue;
 493                         lang.second.readLayoutTranslations(tr.second, m == ExactMatch);
 494                 }
 495         }
 496
 497 }
 498
 499
 500 Language const * Languages::getLanguage(string const & language) const
 501 {
 502         if (language == "reset")
 503                 return reset_language;
 504         if (language == "ignore")
 505                 return ignore_language;
 506         const_iterator it = languagelist_.find(language);
 507         return it == languagelist_.end() ? reset_language : &it->second;
 508 }
 509
 510
 511 } // namespace lyx