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