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