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