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