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 "Lexer.h"
  20 #include "LyXRC.h"
  21
  22 #include "support/debug.h"
  23 #include "support/FileName.h"
  24 #include "support/filetools.h"
  25 #include "support/lassert.h"
  26 #include "support/lstrings.h"
  27 #include "support/Messages.h"
  28
  29 using namespace std;
  30 using namespace lyx::support;
  31
  32 namespace lyx {
  33
  34 Languages languages;
  35 Language const * ignore_language = 0;
  36 Language const * default_language = 0;
  37 Language const * latex_language = 0;
  38 Language const * reset_language = 0;
  39
  40
  41 bool Language::isPolyglossiaExclusive() const
  42 {
  43         return babel().empty() && !polyglossia().empty() && requires().empty();
  44 }
  45
  46
  47 docstring const Language::translateLayout(string const & m) const
  48 {
  49         if (m.empty())
  50                 return docstring();
  51
  52         if (!isAscii(m)) {
  53                 lyxerr << "Warning: not translating `" << m
  54                        << "' because it is not pure ASCII.\n";
  55                 return from_utf8(m);
  56         }
  57
  58         TranslationMap::const_iterator it = layoutTranslations_.find(m);
  59         if (it != layoutTranslations_.end())
  60                 return it->second;
  61
  62         docstring t = from_ascii(m);
  63         cleanTranslation(t);
  64         return t;
  65 }
  66
  67
  68 bool Language::readLanguage(Lexer & lex)
  69 {
  70         enum LanguageTags {
  71                 LA_AS_BABELOPTS = 1,
  72                 LA_BABELNAME,
  73                 LA_ENCODING,
  74                 LA_END,
  75                 LA_FONTENC,
  76                 LA_GUINAME,
  77                 LA_INTERNAL_ENC,
  78                 LA_LANG_CODE,
  79                 LA_LANG_VARIETY,
  80                 LA_POLYGLOSSIANAME,
  81                 LA_POLYGLOSSIAOPTS,
  82                 LA_POSTBABELPREAMBLE,
  83                 LA_QUOTESTYLE,
  84                 LA_PREBABELPREAMBLE,
  85                 LA_REQUIRES,
  86                 LA_RTL
  87         };
  88
  89         // Keep these sorted alphabetically!
  90         LexerKeyword languageTags[] = {
  91                 { "asbabeloptions",       LA_AS_BABELOPTS },
  92                 { "babelname",            LA_BABELNAME },
  93                 { "encoding",             LA_ENCODING },
  94                 { "end",                  LA_END },
  95                 { "fontencoding",         LA_FONTENC },
  96                 { "guiname",              LA_GUINAME },
  97                 { "internalencoding",     LA_INTERNAL_ENC },
  98                 { "langcode",             LA_LANG_CODE },
  99                 { "langvariety",          LA_LANG_VARIETY },
 100                 { "polyglossianame",      LA_POLYGLOSSIANAME },
 101                 { "polyglossiaopts",      LA_POLYGLOSSIAOPTS },
 102                 { "postbabelpreamble",    LA_POSTBABELPREAMBLE },
 103                 { "prebabelpreamble",     LA_PREBABELPREAMBLE },
 104                 { "quotestyle",           LA_QUOTESTYLE },
 105                 { "requires",             LA_REQUIRES },
 106                 { "rtl",                  LA_RTL }
 107         };
 108
 109         bool error = false;
 110         bool finished = false;
 111         lex.pushTable(languageTags);
 112         // parse style section
 113         while (!finished && lex.isOK() && !error) {
 114                 int le = lex.lex();
 115                 // See comment in LyXRC.cpp.
 116                 switch (le) {
 117                 case Lexer::LEX_FEOF:
 118                         continue;
 119
 120                 case Lexer::LEX_UNDEF: // parse error
 121                         lex.printError("Unknown language tag `$$Token'");
 122                         error = true;
 123                         continue;
 124
 125                 default:
 126                         break;
 127                 }
 128                 switch (static_cast<LanguageTags>(le)) {
 129                 case LA_END: // end of structure
 130                         finished = true;
 131                         break;
 132                 case LA_AS_BABELOPTS:
 133                         lex >> as_babel_options_;
 134                         break;
 135                 case LA_BABELNAME:
 136                         lex >> babel_;
 137                         break;
 138                 case LA_POLYGLOSSIANAME:
 139                         lex >> polyglossia_name_;
 140                         break;
 141                 case LA_POLYGLOSSIAOPTS:
 142                         lex >> polyglossia_opts_;
 143                         break;
 144                 case LA_QUOTESTYLE:
 145                         lex >> quote_style_;
 146                         break;
 147                 case LA_ENCODING:
 148                         lex >> encodingStr_;
 149                         break;
 150                 case LA_FONTENC:
 151                         lex >> fontenc_;
 152                         break;
 153                 case LA_GUINAME:
 154                         lex >> display_;
 155                         break;
 156                 case LA_INTERNAL_ENC:
 157                         lex >> internal_enc_;
 158                         break;
 159                 case LA_LANG_CODE:
 160                         lex >> code_;
 161                         break;
 162                 case LA_LANG_VARIETY:
 163                         lex >> variety_;
 164                         break;
 165                 case LA_POSTBABELPREAMBLE:
 166                         babel_postsettings_ =
 167                                 lex.getLongString("EndPostBabelPreamble");
 168                         break;
 169                 case LA_PREBABELPREAMBLE:
 170                         babel_presettings_ =
 171                                 lex.getLongString("EndPreBabelPreamble");
 172                         break;
 173                 case LA_REQUIRES:
 174                         lex >> requires_;
 175                         break;
 176                 case LA_RTL:
 177                         lex >> rightToLeft_;
 178                         break;
 179                 }
 180         }
 181         lex.popTable();
 182         return finished && !error;
 183 }
 184
 185
 186 bool Language::read(Lexer & lex)
 187 {
 188         as_babel_options_ = 0;
 189         encoding_ = 0;
 190         internal_enc_ = 0;
 191         rightToLeft_ = 0;
 192
 193         if (!lex.next()) {
 194                 lex.printError("No name given for language: `$$Token'.");
 195                 return false;
 196         }
 197
 198         lang_ = lex.getString();
 199         LYXERR(Debug::INFO, "Reading language " << lang_);
 200         if (!readLanguage(lex)) {
 201                 LYXERR0("Error parsing language `" << lang_ << '\'');
 202                 return false;
 203         }
 204
 205         encoding_ = encodings.fromLyXName(encodingStr_);
 206         if (!encoding_ && !encodingStr_.empty()) {
 207                 encoding_ = encodings.fromLyXName("iso8859-1");
 208                 LYXERR0("Unknown encoding " << encodingStr_);
 209         }
 210         return true;
 211 }
 212
 213
 214 void Language::readLayoutTranslations(Language::TranslationMap const & trans, bool replace)
 215 {
 216         TranslationMap::const_iterator const end = trans.end();
 217         for (TranslationMap::const_iterator it = trans.begin(); it != end; ++it) {
 218                 if (replace
 219                         || layoutTranslations_.find(it->first) == layoutTranslations_.end())
 220                         layoutTranslations_[it->first] = it->second;
 221         }
 222 }
 223
 224
 225 void Languages::read(FileName const & filename)
 226 {
 227         Lexer lex;
 228         lex.setFile(filename);
 229         lex.setContext("Languages::read");
 230         while (lex.isOK()) {
 231                 int le = lex.lex();
 232                 switch (le) {
 233                 case Lexer::LEX_FEOF:
 234                         continue;
 235
 236                 default:
 237                         break;
 238                 }
 239                 if (lex.getString() != "Language") {
 240                         lex.printError("Unknown Language tag `$$Token'");
 241                         continue;
 242                 }
 243                 Language l;
 244                 l.read(lex);
 245                 if (!lex)
 246                         break;
 247                 if (l.lang() == "latex") {
 248                         // Check if latex language was not already defined.
 249                         LASSERT(latex_language == 0, continue);
 250                         static const Language latex_lang = l;
 251                         latex_language = &latex_lang;
 252                 } else if (l.lang() == "ignore") {
 253                         // Check if ignore language was not already defined.
 254                         LASSERT(ignore_language == 0, continue);
 255                         static const Language ignore_lang = l;
 256                         ignore_language = &ignore_lang;
 257                 } else
 258                         languagelist[l.lang()] = l;
 259         }
 260
 261         default_language = getLanguage("english");
 262         if (!default_language) {
 263                 LYXERR0("Default language \"english\" not found!");
 264                 default_language = &(*languagelist.begin()).second;
 265                 LYXERR0("Using \"" << default_language->lang() << "\" instead!");
 266         }
 267
 268         // Read layout translations
 269         FileName const path = libFileSearch(string(), "layouttranslations");
 270         readLayoutTranslations(path);
 271 }
 272
 273
 274 namespace {
 275
 276 bool readTranslations(Lexer & lex, Language::TranslationMap & trans)
 277 {
 278         while (lex.isOK()) {
 279                 if (lex.checkFor("End"))
 280                         break;
 281                 if (!lex.next(true))
 282                         return false;
 283                 string const key = lex.getString();
 284                 if (!lex.next(true))
 285                         return false;
 286                 docstring const val = lex.getDocString();
 287                 trans[key] = val;
 288         }
 289         return true;
 290 }
 291
 292
 293 enum Match {
 294         NoMatch,
 295         ApproximateMatch,
 296         ExactMatch
 297 };
 298
 299
 300 Match match(string const & code, Language const & lang)
 301 {
 302         // we need to mimic gettext: code can be a two-letter code, which
 303         // should match all variants, e.g. "de" should match "de_DE",
 304         // "de_AT" etc.
 305         // special case for chinese:
 306         // simplified  => code == "zh_CN", langcode == "zh_CN"
 307         // traditional => code == "zh_TW", langcode == "zh_CN"
 308         string const variety = lang.variety();
 309         string const langcode = variety.empty() ?
 310                                 lang.code() : lang.code() + '_' + variety;
 311         string const name = lang.lang();
 312         if ((code == langcode && name != "chinese-traditional")
 313                 || (code == "zh_TW"  && name == "chinese-traditional"))
 314                 return ExactMatch;
 315         if ((code.size() == 2) && (langcode.size() > 2)
 316                 && (code + '_' == langcode.substr(0, 3)))
 317                 return ApproximateMatch;
 318         return NoMatch;
 319 }
 320
 321 }
 322
 323
 324 void Languages::readLayoutTranslations(support::FileName const & filename)
 325 {
 326         Lexer lex;
 327         lex.setFile(filename);
 328         lex.setContext("Languages::read");
 329
 330         // 1) read all translations (exact and approximate matches) into trans
 331         typedef std::map<string, Language::TranslationMap> TransMap;
 332         TransMap trans;
 333         LanguageList::iterator const lbeg = languagelist.begin();
 334         LanguageList::iterator const lend = languagelist.end();
 335         while (lex.isOK()) {
 336                 if (!lex.checkFor("Translation")) {
 337                         if (lex.isOK())
 338                                 lex.printError("Unknown layout translation tag `$$Token'");
 339                         break;
 340                 }
 341                 if (!lex.next(true))
 342                         break;
 343                 string const code = lex.getString();
 344                 bool found = false;
 345                 for (LanguageList::iterator lit = lbeg; lit != lend; ++lit) {
 346                         if (match(code, lit->second) != NoMatch) {
 347                                 found = true;
 348                                 break;
 349                         }
 350                 }
 351                 if (!found) {
 352                         lex.printError("Unknown language `" + code + "'");
 353                         break;
 354                 }
 355                 if (!readTranslations(lex, trans[code])) {
 356                         lex.printError("Could not read layout translations for language `"
 357                                 + code + "'");
 358                         break;
 359                 }
 360         }
 361
 362         // 2) merge all translations into the languages
 363         // exact translations overwrite approximate ones
 364         TransMap::const_iterator const tbeg = trans.begin();
 365         TransMap::const_iterator const tend = trans.end();
 366         for (TransMap::const_iterator tit = tbeg; tit != tend; ++tit) {
 367                 for (LanguageList::iterator lit = lbeg; lit != lend; ++lit) {
 368                         Match const m = match(tit->first, lit->second);
 369                         if (m == NoMatch)
 370                                 continue;
 371                         lit->second.readLayoutTranslations(tit->second,
 372                                                            m == ExactMatch);
 373                 }
 374         }
 375
 376 }
 377
 378
 379 Language const * Languages::getLanguage(string const & language) const
 380 {
 381         if (language == "reset")
 382                 return reset_language;
 383         if (language == "ignore")
 384                 return ignore_language;
 385         const_iterator it = languagelist.find(language);
 386         return it == languagelist.end() ? reset_language : &it->second;
 387 }
 388
 389
 390 } // namespace lyx