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