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                 LA_ACTIVECHARS
 151         };
 152
 153         // Keep these sorted alphabetically!
 154         LexerKeyword languageTags[] = {
 155                 { "activechars",          LA_ACTIVECHARS },
 156                 { "babelname",            LA_BABELNAME },
 157                 { "dateformats",          LA_DATEFORMATS },
 158                 { "encoding",             LA_ENCODING },
 159                 { "end",                  LA_END },
 160                 { "fontencoding",         LA_FONTENC },
 161                 { "guiname",              LA_GUINAME },
 162                 { "hasguisupport",        LA_HAS_GUI_SUPPORT },
 163                 { "internalencoding",     LA_INTERNAL_ENC },
 164                 { "langcode",             LA_LANG_CODE },
 165                 { "langvariety",          LA_LANG_VARIETY },
 166                 { "polyglossianame",      LA_POLYGLOSSIANAME },
 167                 { "polyglossiaopts",      LA_POLYGLOSSIAOPTS },
 168                 { "postbabelpreamble",    LA_POSTBABELPREAMBLE },
 169                 { "prebabelpreamble",     LA_PREBABELPREAMBLE },
 170                 { "provides",             LA_PROVIDES },
 171                 { "quotestyle",           LA_QUOTESTYLE },
 172                 { "requires",             LA_REQUIRES },
 173                 { "rtl",                  LA_RTL },
 174                 {"wordwrap",              LA_WORDWRAP }
 175         };
 176
 177         bool error = false;
 178         bool finished = false;
 179         lex.pushTable(languageTags);
 180         // parse style section
 181         while (!finished && lex.isOK() && !error) {
 182                 int le = lex.lex();
 183                 // See comment in LyXRC.cpp.
 184                 switch (le) {
 185                 case Lexer::LEX_FEOF:
 186                         continue;
 187
 188                 case Lexer::LEX_UNDEF: // parse error
 189                         lex.printError("Unknown language tag `$$Token'");
 190                         error = true;
 191                         continue;
 192
 193                 default:
 194                         break;
 195                 }
 196                 switch (static_cast<LanguageTags>(le)) {
 197                 case LA_END: // end of structure
 198                         finished = true;
 199                         break;
 200                 case LA_BABELNAME:
 201                         lex >> babel_;
 202                         break;
 203                 case LA_POLYGLOSSIANAME:
 204                         lex >> polyglossia_name_;
 205                         break;
 206                 case LA_POLYGLOSSIAOPTS:
 207                         lex >> polyglossia_opts_;
 208                         break;
 209                 case LA_QUOTESTYLE:
 210                         lex >> quote_style_;
 211                         break;
 212                 case LA_ACTIVECHARS:
 213                         lex >> active_chars_;
 214                         break;
 215                 case LA_ENCODING:
 216                         lex >> encodingStr_;
 217                         break;
 218                 case LA_FONTENC: {
 219                         lex.eatLine();
 220                         vector<string> const fe =
 221                                 getVectorFromString(lex.getString(true), "|");
 222                         fontenc_.insert(fontenc_.end(), fe.begin(), fe.end());
 223                         break;
 224                 }
 225                 case LA_DATEFORMATS: {
 226                         lex.eatLine();
 227                         vector<string> const df =
 228                                 getVectorFromString(trim(lex.getString(true), "\""), "|");
 229                         dateformats_.insert(dateformats_.end(), df.begin(), df.end());
 230                         break;
 231                 }
 232                 case LA_GUINAME:
 233                         lex >> display_;
 234                         break;
 235                 case LA_HAS_GUI_SUPPORT:
 236                         lex >> has_gui_support_;
 237                         break;
 238                 case LA_INTERNAL_ENC:
 239                         lex >> internal_enc_;
 240                         break;
 241                 case LA_LANG_CODE:
 242                         lex >> code_;
 243                         break;
 244                 case LA_LANG_VARIETY:
 245                         lex >> variety_;
 246                         break;
 247                 case LA_POSTBABELPREAMBLE:
 248                         babel_postsettings_ =
 249                                 lex.getLongString(from_ascii("EndPostBabelPreamble"));
 250                         break;
 251                 case LA_PREBABELPREAMBLE:
 252                         babel_presettings_ =
 253                                 lex.getLongString(from_ascii("EndPreBabelPreamble"));
 254                         break;
 255                 case LA_REQUIRES:
 256                         lex >> requires_;
 257                         break;
 258                 case LA_PROVIDES:
 259                         lex >> provides_;
 260                         break;
 261                 case LA_RTL:
 262                         lex >> rightToLeft_;
 263                         break;
 264                 case LA_WORDWRAP:
 265                         lex >> word_wrap_;
 266                         break;
 267                 }
 268         }
 269         lex.popTable();
 270         return finished && !error;
 271 }
 272
 273
 274 bool Language::read(Lexer & lex)
 275 {
 276         encoding_ = 0;
 277         internal_enc_ = 0;
 278         rightToLeft_ = 0;
 279
 280         if (!lex.next()) {
 281                 lex.printError("No name given for language: `$$Token'.");
 282                 return false;
 283         }
 284
 285         lang_ = lex.getString();
 286         LYXERR(Debug::INFO, "Reading language " << lang_);
 287         if (!readLanguage(lex)) {
 288                 LYXERR0("Error parsing language `" << lang_ << '\'');
 289                 return false;
 290         }
 291
 292         encoding_ = encodings.fromLyXName(encodingStr_);
 293         if (!encoding_ && !encodingStr_.empty()) {
 294                 encoding_ = encodings.fromLyXName("iso8859-1");
 295                 LYXERR0("Unknown encoding " << encodingStr_);
 296         }
 297         if (fontenc_.empty())
 298                 fontenc_.push_back("ASCII");
 299         if (dateformats_.empty()) {
 300                 dateformats_.push_back("MMMM dd, yyyy");
 301                 dateformats_.push_back("MMM dd, yyyy");
 302                 dateformats_.push_back("M/d/yyyy");
 303         }
 304         return true;
 305 }
 306
 307
 308 void Language::readLayoutTranslations(Language::TranslationMap const & trans, bool replace)
 309 {
 310         for (auto const & t : trans) {
 311                 if (replace
 312                     || layoutTranslations_.find(t.first) == layoutTranslations_.end())
 313                         layoutTranslations_[t.first] = t.second;
 314         }
 315 }
 316
 317
 318 void Languages::read(FileName const & filename)
 319 {
 320         Lexer lex;
 321         lex.setFile(filename);
 322         lex.setContext("Languages::read");
 323         while (lex.isOK()) {
 324                 int le = lex.lex();
 325                 switch (le) {
 326                 case Lexer::LEX_FEOF:
 327                         continue;
 328
 329                 default:
 330                         break;
 331                 }
 332                 if (lex.getString() != "Language") {
 333                         lex.printError("Unknown Language tag `$$Token'");
 334                         continue;
 335                 }
 336                 Language l;
 337                 l.read(lex);
 338                 if (!lex)
 339                         break;
 340                 if (l.lang() == "latex") {
 341                         // Check if latex language was not already defined.
 342                         LASSERT(latex_language == 0, continue);
 343                         static const Language latex_lang = l;
 344                         latex_language = &latex_lang;
 345                 } else if (l.lang() == "ignore") {
 346                         // Check if ignore language was not already defined.
 347                         LASSERT(ignore_language == 0, continue);
 348                         static const Language ignore_lang = l;
 349                         ignore_language = &ignore_lang;
 350                 } else
 351                         languagelist_[l.lang()] = l;
 352         }
 353
 354         default_language = getLanguage("english");
 355         if (!default_language) {
 356                 LYXERR0("Default language \"english\" not found!");
 357                 default_language = &(*languagelist_.begin()).second;
 358                 LYXERR0("Using \"" << default_language->lang() << "\" instead!");
 359         }
 360
 361         // Read layout translations
 362         FileName const path = libFileSearch(string(), "layouttranslations");
 363         readLayoutTranslations(path);
 364 }
 365
 366
 367 namespace {
 368
 369 bool readTranslations(Lexer & lex, Language::TranslationMap & trans)
 370 {
 371         while (lex.isOK()) {
 372                 if (lex.checkFor("End"))
 373                         break;
 374                 if (!lex.next(true))
 375                         return false;
 376                 string const key = lex.getString();
 377                 if (!lex.next(true))
 378                         return false;
 379                 docstring const val = lex.getDocString();
 380                 trans[key] = val;
 381         }
 382         return true;
 383 }
 384
 385
 386 enum Match {
 387         NoMatch,
 388         ApproximateMatch,
 389         VeryApproximateMatch,
 390         ExactMatch
 391 };
 392
 393
 394 Match match(string const & code, Language const & lang)
 395 {
 396         // we need to mimic gettext: code can be a two-letter code, which
 397         // should match all variants, e.g. "de" should match "de_DE",
 398         // "de_AT" etc.
 399         // special case for chinese:
 400         // simplified  => code == "zh_CN", langcode == "zh_CN"
 401         // traditional => code == "zh_TW", langcode == "zh_CN"
 402         string const variety = lang.variety();
 403         string const langcode = variety.empty() ?
 404                                 lang.code() : lang.code() + '_' + variety;
 405         string const name = lang.lang();
 406         if ((code == langcode && name != "chinese-traditional")
 407                 || (code == "zh_TW"  && name == "chinese-traditional"))
 408                 return ExactMatch;
 409         if ((code.size() == 2) && (langcode.size() > 2)
 410                 && (code + '_' == langcode.substr(0, 3)))
 411                 return ApproximateMatch;
 412         if (code.substr(0,2) == langcode.substr(0,2))
 413                 return VeryApproximateMatch;
 414         return NoMatch;
 415 }
 416
 417 } // namespace
 418
 419
 420
 421 Language const * Languages::getFromCode(string const & code) const
 422 {
 423         // 1/ exact match with any known language
 424         for (auto const & l : languagelist_) {
 425                 if (match(code, l.second) == ExactMatch)
 426                         return &l.second;
 427         }
 428
 429         // 2/ approximate with any known language
 430         for (auto const & l : languagelist_) {
 431                 if (match(code, l.second) == ApproximateMatch)
 432                         return &l.second;
 433         }
 434         return 0;
 435 }
 436
 437
 438 Language const * Languages::getFromCode(string const & code,
 439                         set<Language const *> const & tryfirst) const
 440 {
 441         // 1/ exact match with tryfirst list
 442         for (auto const * lptr : tryfirst) {
 443                 if (match(code, *lptr) == ExactMatch)
 444                         return lptr;
 445         }
 446
 447         // 2/ approximate match with tryfirst list
 448         for (auto const * lptr : tryfirst) {
 449                 Match const m = match(code, *lptr);
 450                 if (m == ApproximateMatch || m == VeryApproximateMatch)
 451                         return lptr;
 452         }
 453
 454         // 3/ stricter match in all languages
 455         return getFromCode(code);
 456
 457         LYXERR0("Unknown language `" << code << "'");
 458         return 0;
 459 }
 460
 461
 462 void Languages::readLayoutTranslations(support::FileName const & filename)
 463 {
 464         Lexer lex;
 465         lex.setFile(filename);
 466         lex.setContext("Languages::read");
 467
 468         // 1) read all translations (exact and approximate matches) into trans
 469         std::map<string, Language::TranslationMap> trans;
 470         while (lex.isOK()) {
 471                 if (!lex.checkFor("Translation")) {
 472                         if (lex.isOK())
 473                                 lex.printError("Unknown layout translation tag `$$Token'");
 474                         break;
 475                 }
 476                 if (!lex.next(true))
 477                         break;
 478                 string const code = lex.getString();
 479                 bool found = getFromCode(code);
 480                 if (!found) {
 481                         lex.printError("Unknown language `" + code + "'");
 482                         break;
 483                 }
 484                 if (!readTranslations(lex, trans[code])) {
 485                         lex.printError("Could not read layout translations for language `"
 486                                 + code + "'");
 487                         break;
 488                 }
 489         }
 490
 491         // 2) merge all translations into the languages
 492         // exact translations overwrite approximate ones
 493         for (auto & tr : trans) {
 494                 for (auto & lang : languagelist_) {
 495                         Match const m = match(tr.first, lang.second);
 496                         if (m == NoMatch)
 497                                 continue;
 498                         lang.second.readLayoutTranslations(tr.second, m == ExactMatch);
 499                 }
 500         }
 501
 502 }
 503
 504
 505 Language const * Languages::getLanguage(string const & language) const
 506 {
 507         if (language == "reset")
 508                 return reset_language;
 509         if (language == "ignore")
 510                 return ignore_language;
 511         const_iterator it = languagelist_.find(language);
 512         return it == languagelist_.end() ? reset_language : &it->second;
 513 }
 514
 515
 516 } // namespace lyx