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