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