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/Messages.h"
  29
  30 using namespace std;
  31 using namespace lyx::support;
  32
  33 namespace lyx {
  34
  35 Languages languages;
  36 Language const * ignore_language = 0;
  37 Language const * default_language = 0;
  38 Language const * latex_language = 0;
  39 Language const * reset_language = 0;
  40
  41
  42 bool Language::isPolyglossiaExclusive() const
  43 {
  44         return babel().empty() && !polyglossia().empty() && requires().empty();
  45 }
  46
  47
  48 bool Language::isBabelExclusive() const
  49 {
  50         return !babel().empty() && polyglossia().empty() && requires().empty();
  51 }
  52
  53
  54 docstring const Language::translateLayout(string const & m) const
  55 {
  56         if (m.empty())
  57                 return docstring();
  58
  59         if (!isAscii(m)) {
  60                 lyxerr << "Warning: not translating `" << m
  61                        << "' because it is not pure ASCII.\n";
  62                 return from_utf8(m);
  63         }
  64
  65         TranslationMap::const_iterator it = layoutTranslations_.find(m);
  66         if (it != layoutTranslations_.end())
  67                 return it->second;
  68
  69         docstring t = from_ascii(m);
  70         cleanTranslation(t);
  71         return t;
  72 }
  73
  74
  75 vector<string> Language::fontencs() const
  76 {
  77         return fontenc_;
  78 }
  79
  80
  81 string Language::fontenc(BufferParams const & params) const
  82 {
  83         // Determine optimal font encoding
  84         // We check whether the used rm font supports an encoding our language supports
  85         for (auto & fe : fontenc_) {
  86                 LaTeXFont const & lf = theLaTeXFonts().getLaTeXFont(from_ascii(params.fontsRoman()));
  87                 // ASCII means: support all T* encodings plus OT1
  88                 if (fe == "ASCII") {
  89                         vector<string> const lfe = lf.fontencs();
  90                         for (auto & afe : lfe) {
  91                                 if (afe == "OT1" || prefixIs(afe, "T"))
  92                                         // we found a suitable one; return that.
  93                                         return afe;
  94                         }
  95                 }
  96                 // For other encodings, just check whether the font supports it
  97                 if (lf.hasFontenc(fe))
  98                         return fe;
  99         }
 100         // We did not find a suitable one; just take the first in the list,
 101         // the priorized one (which is "T1" for ASCII).
 102         return fontencs().front() == "ASCII" ? "T1" : fontencs().front();
 103 }
 104
 105
 106 bool Language::readLanguage(Lexer & lex)
 107 {
 108         enum LanguageTags {
 109                 LA_AS_BABELOPTS = 1,
 110                 LA_BABELNAME,
 111                 LA_ENCODING,
 112                 LA_END,
 113                 LA_FONTENC,
 114                 LA_GUINAME,
 115                 LA_HAS_GUI_SUPPORT,
 116                 LA_INTERNAL_ENC,
 117                 LA_LANG_CODE,
 118                 LA_LANG_VARIETY,
 119                 LA_POLYGLOSSIANAME,
 120                 LA_POLYGLOSSIAOPTS,
 121                 LA_POSTBABELPREAMBLE,
 122                 LA_PREBABELPREAMBLE,
 123                 LA_PROVIDES,
 124                 LA_REQUIRES,
 125                 LA_QUOTESTYLE,
 126                 LA_RTL
 127         };
 128
 129         // Keep these sorted alphabetically!
 130         LexerKeyword languageTags[] = {
 131                 { "asbabeloptions",       LA_AS_BABELOPTS },
 132                 { "babelname",            LA_BABELNAME },
 133                 { "encoding",             LA_ENCODING },
 134                 { "end",                  LA_END },
 135                 { "fontencoding",         LA_FONTENC },
 136                 { "guiname",              LA_GUINAME },
 137                 { "hasguisupport",        LA_HAS_GUI_SUPPORT },
 138                 { "internalencoding",     LA_INTERNAL_ENC },
 139                 { "langcode",             LA_LANG_CODE },
 140                 { "langvariety",          LA_LANG_VARIETY },
 141                 { "polyglossianame",      LA_POLYGLOSSIANAME },
 142                 { "polyglossiaopts",      LA_POLYGLOSSIAOPTS },
 143                 { "postbabelpreamble",    LA_POSTBABELPREAMBLE },
 144                 { "prebabelpreamble",     LA_PREBABELPREAMBLE },
 145                 { "provides",             LA_PROVIDES },
 146                 { "quotestyle",           LA_QUOTESTYLE },
 147                 { "requires",             LA_REQUIRES },
 148                 { "rtl",                  LA_RTL }
 149         };
 150
 151         bool error = false;
 152         bool finished = false;
 153         lex.pushTable(languageTags);
 154         // parse style section
 155         while (!finished && lex.isOK() && !error) {
 156                 int le = lex.lex();
 157                 // See comment in LyXRC.cpp.
 158                 switch (le) {
 159                 case Lexer::LEX_FEOF:
 160                         continue;
 161
 162                 case Lexer::LEX_UNDEF: // parse error
 163                         lex.printError("Unknown language tag `$$Token'");
 164                         error = true;
 165                         continue;
 166
 167                 default:
 168                         break;
 169                 }
 170                 switch (static_cast<LanguageTags>(le)) {
 171                 case LA_END: // end of structure
 172                         finished = true;
 173                         break;
 174                 case LA_AS_BABELOPTS:
 175                         lex >> as_babel_options_;
 176                         break;
 177                 case LA_BABELNAME:
 178                         lex >> babel_;
 179                         break;
 180                 case LA_POLYGLOSSIANAME:
 181                         lex >> polyglossia_name_;
 182                         break;
 183                 case LA_POLYGLOSSIAOPTS:
 184                         lex >> polyglossia_opts_;
 185                         break;
 186                 case LA_QUOTESTYLE:
 187                         lex >> quote_style_;
 188                         break;
 189                 case LA_ENCODING:
 190                         lex >> encodingStr_;
 191                         break;
 192                 case LA_FONTENC: {
 193                         lex.eatLine();
 194                         vector<string> const fe =
 195                                 getVectorFromString(lex.getString(true), "|");
 196                         fontenc_.insert(fontenc_.end(), fe.begin(), fe.end());
 197                         break;
 198                 }
 199                 case LA_GUINAME:
 200                         lex >> display_;
 201                         break;
 202                 case LA_HAS_GUI_SUPPORT:
 203                         lex >> has_gui_support_;
 204                         break;
 205                 case LA_INTERNAL_ENC:
 206                         lex >> internal_enc_;
 207                         break;
 208                 case LA_LANG_CODE:
 209                         lex >> code_;
 210                         break;
 211                 case LA_LANG_VARIETY:
 212                         lex >> variety_;
 213                         break;
 214                 case LA_POSTBABELPREAMBLE:
 215                         babel_postsettings_ =
 216                                 lex.getLongString(from_ascii("EndPostBabelPreamble"));
 217                         break;
 218                 case LA_PREBABELPREAMBLE:
 219                         babel_presettings_ =
 220                                 lex.getLongString(from_ascii("EndPreBabelPreamble"));
 221                         break;
 222                 case LA_REQUIRES:
 223                         lex >> requires_;
 224                         break;
 225                 case LA_PROVIDES:
 226                         lex >> provides_;
 227                         break;
 228                 case LA_RTL:
 229                         lex >> rightToLeft_;
 230                         break;
 231                 }
 232         }
 233         lex.popTable();
 234         return finished && !error;
 235 }
 236
 237
 238 bool Language::read(Lexer & lex)
 239 {
 240         as_babel_options_ = 0;
 241         encoding_ = 0;
 242         internal_enc_ = 0;
 243         rightToLeft_ = 0;
 244
 245         if (!lex.next()) {
 246                 lex.printError("No name given for language: `$$Token'.");
 247                 return false;
 248         }
 249
 250         lang_ = lex.getString();
 251         LYXERR(Debug::INFO, "Reading language " << lang_);
 252         if (!readLanguage(lex)) {
 253                 LYXERR0("Error parsing language `" << lang_ << '\'');
 254                 return false;
 255         }
 256
 257         encoding_ = encodings.fromLyXName(encodingStr_);
 258         if (!encoding_ && !encodingStr_.empty()) {
 259                 encoding_ = encodings.fromLyXName("iso8859-1");
 260                 LYXERR0("Unknown encoding " << encodingStr_);
 261         }
 262         if (fontenc_.empty())
 263                 fontenc_.push_back("ASCII");
 264         return true;
 265 }
 266
 267
 268 void Language::readLayoutTranslations(Language::TranslationMap const & trans, bool replace)
 269 {
 270         TranslationMap::const_iterator const end = trans.end();
 271         for (TranslationMap::const_iterator it = trans.begin(); it != end; ++it) {
 272                 if (replace
 273                         || layoutTranslations_.find(it->first) == layoutTranslations_.end())
 274                         layoutTranslations_[it->first] = it->second;
 275         }
 276 }
 277
 278
 279 void Languages::read(FileName const & filename)
 280 {
 281         Lexer lex;
 282         lex.setFile(filename);
 283         lex.setContext("Languages::read");
 284         while (lex.isOK()) {
 285                 int le = lex.lex();
 286                 switch (le) {
 287                 case Lexer::LEX_FEOF:
 288                         continue;
 289
 290                 default:
 291                         break;
 292                 }
 293                 if (lex.getString() != "Language") {
 294                         lex.printError("Unknown Language tag `$$Token'");
 295                         continue;
 296                 }
 297                 Language l;
 298                 l.read(lex);
 299                 if (!lex)
 300                         break;
 301                 if (l.lang() == "latex") {
 302                         // Check if latex language was not already defined.
 303                         LASSERT(latex_language == 0, continue);
 304                         static const Language latex_lang = l;
 305                         latex_language = &latex_lang;
 306                 } else if (l.lang() == "ignore") {
 307                         // Check if ignore language was not already defined.
 308                         LASSERT(ignore_language == 0, continue);
 309                         static const Language ignore_lang = l;
 310                         ignore_language = &ignore_lang;
 311                 } else
 312                         languagelist[l.lang()] = l;
 313         }
 314
 315         default_language = getLanguage("english");
 316         if (!default_language) {
 317                 LYXERR0("Default language \"english\" not found!");
 318                 default_language = &(*languagelist.begin()).second;
 319                 LYXERR0("Using \"" << default_language->lang() << "\" instead!");
 320         }
 321
 322         // Read layout translations
 323         FileName const path = libFileSearch(string(), "layouttranslations");
 324         readLayoutTranslations(path);
 325 }
 326
 327
 328 namespace {
 329
 330 bool readTranslations(Lexer & lex, Language::TranslationMap & trans)
 331 {
 332         while (lex.isOK()) {
 333                 if (lex.checkFor("End"))
 334                         break;
 335                 if (!lex.next(true))
 336                         return false;
 337                 string const key = lex.getString();
 338                 if (!lex.next(true))
 339                         return false;
 340                 docstring const val = lex.getDocString();
 341                 trans[key] = val;
 342         }
 343         return true;
 344 }
 345
 346
 347 enum Match {
 348         NoMatch,
 349         ApproximateMatch,
 350         ExactMatch
 351 };
 352
 353
 354 Match match(string const & code, Language const & lang)
 355 {
 356         // we need to mimic gettext: code can be a two-letter code, which
 357         // should match all variants, e.g. "de" should match "de_DE",
 358         // "de_AT" etc.
 359         // special case for chinese:
 360         // simplified  => code == "zh_CN", langcode == "zh_CN"
 361         // traditional => code == "zh_TW", langcode == "zh_CN"
 362         string const variety = lang.variety();
 363         string const langcode = variety.empty() ?
 364                                 lang.code() : lang.code() + '_' + variety;
 365         string const name = lang.lang();
 366         if ((code == langcode && name != "chinese-traditional")
 367                 || (code == "zh_TW"  && name == "chinese-traditional"))
 368                 return ExactMatch;
 369         if ((code.size() == 2) && (langcode.size() > 2)
 370                 && (code + '_' == langcode.substr(0, 3)))
 371                 return ApproximateMatch;
 372         return NoMatch;
 373 }
 374
 375 } // namespace
 376
 377
 378 void Languages::readLayoutTranslations(support::FileName const & filename)
 379 {
 380         Lexer lex;
 381         lex.setFile(filename);
 382         lex.setContext("Languages::read");
 383
 384         // 1) read all translations (exact and approximate matches) into trans
 385         typedef std::map<string, Language::TranslationMap> TransMap;
 386         TransMap trans;
 387         LanguageList::iterator const lbeg = languagelist.begin();
 388         LanguageList::iterator const lend = languagelist.end();
 389         while (lex.isOK()) {
 390                 if (!lex.checkFor("Translation")) {
 391                         if (lex.isOK())
 392                                 lex.printError("Unknown layout translation tag `$$Token'");
 393                         break;
 394                 }
 395                 if (!lex.next(true))
 396                         break;
 397                 string const code = lex.getString();
 398                 bool found = false;
 399                 for (LanguageList::iterator lit = lbeg; lit != lend; ++lit) {
 400                         if (match(code, lit->second) != NoMatch) {
 401                                 found = true;
 402                                 break;
 403                         }
 404                 }
 405                 if (!found) {
 406                         lex.printError("Unknown language `" + code + "'");
 407                         break;
 408                 }
 409                 if (!readTranslations(lex, trans[code])) {
 410                         lex.printError("Could not read layout translations for language `"
 411                                 + code + "'");
 412                         break;
 413                 }
 414         }
 415
 416         // 2) merge all translations into the languages
 417         // exact translations overwrite approximate ones
 418         TransMap::const_iterator const tbeg = trans.begin();
 419         TransMap::const_iterator const tend = trans.end();
 420         for (TransMap::const_iterator tit = tbeg; tit != tend; ++tit) {
 421                 for (LanguageList::iterator lit = lbeg; lit != lend; ++lit) {
 422                         Match const m = match(tit->first, lit->second);
 423                         if (m == NoMatch)
 424                                 continue;
 425                         lit->second.readLayoutTranslations(tit->second,
 426                                                            m == ExactMatch);
 427                 }
 428         }
 429
 430 }
 431
 432
 433 Language const * Languages::getLanguage(string const & language) const
 434 {
 435         if (language == "reset")
 436                 return reset_language;
 437         if (language == "ignore")
 438                 return ignore_language;
 439         const_iterator it = languagelist.find(language);
 440         return it == languagelist.end() ? reset_language : &it->second;
 441 }
 442
 443
 444 } // namespace lyx