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 bool Language::isBabelExclusive() const
  48 {
  49         return !babel().empty() && polyglossia().empty() && requires().empty();
  50 }
  51
  52
  53 docstring const Language::translateLayout(string const & m) const
  54 {
  55         if (m.empty())
  56                 return docstring();
  57
  58         if (!isAscii(m)) {
  59                 lyxerr << "Warning: not translating `" << m
  60                        << "' because it is not pure ASCII.\n";
  61                 return from_utf8(m);
  62         }
  63
  64         TranslationMap::const_iterator it = layoutTranslations_.find(m);
  65         if (it != layoutTranslations_.end())
  66                 return it->second;
  67
  68         docstring t = from_ascii(m);
  69         cleanTranslation(t);
  70         return t;
  71 }
  72
  73
  74 bool Language::readLanguage(Lexer & lex)
  75 {
  76         enum LanguageTags {
  77                 LA_AS_BABELOPTS = 1,
  78                 LA_BABELNAME,
  79                 LA_ENCODING,
  80                 LA_END,
  81                 LA_FONTENC,
  82                 LA_GUINAME,
  83                 LA_HAS_GUI_SUPPORT,
  84                 LA_INTERNAL_ENC,
  85                 LA_LANG_CODE,
  86                 LA_LANG_VARIETY,
  87                 LA_POLYGLOSSIANAME,
  88                 LA_POLYGLOSSIAOPTS,
  89                 LA_POSTBABELPREAMBLE,
  90                 LA_PREBABELPREAMBLE,
  91                 LA_PROVIDES,
  92                 LA_REQUIRES,
  93                 LA_QUOTESTYLE,
  94                 LA_RTL
  95         };
  96
  97         // Keep these sorted alphabetically!
  98         LexerKeyword languageTags[] = {
  99                 { "asbabeloptions",       LA_AS_BABELOPTS },
 100                 { "babelname",            LA_BABELNAME },
 101                 { "encoding",             LA_ENCODING },
 102                 { "end",                  LA_END },
 103                 { "fontencoding",         LA_FONTENC },
 104                 { "guiname",              LA_GUINAME },
 105                 { "hasguisupport",        LA_HAS_GUI_SUPPORT },
 106                 { "internalencoding",     LA_INTERNAL_ENC },
 107                 { "langcode",             LA_LANG_CODE },
 108                 { "langvariety",          LA_LANG_VARIETY },
 109                 { "polyglossianame",      LA_POLYGLOSSIANAME },
 110                 { "polyglossiaopts",      LA_POLYGLOSSIAOPTS },
 111                 { "postbabelpreamble",    LA_POSTBABELPREAMBLE },
 112                 { "prebabelpreamble",     LA_PREBABELPREAMBLE },
 113                 { "provides",             LA_PROVIDES },
 114                 { "quotestyle",           LA_QUOTESTYLE },
 115                 { "requires",             LA_REQUIRES },
 116                 { "rtl",                  LA_RTL }
 117         };
 118
 119         bool error = false;
 120         bool finished = false;
 121         lex.pushTable(languageTags);
 122         // parse style section
 123         while (!finished && lex.isOK() && !error) {
 124                 int le = lex.lex();
 125                 // See comment in LyXRC.cpp.
 126                 switch (le) {
 127                 case Lexer::LEX_FEOF:
 128                         continue;
 129
 130                 case Lexer::LEX_UNDEF: // parse error
 131                         lex.printError("Unknown language tag `$$Token'");
 132                         error = true;
 133                         continue;
 134
 135                 default:
 136                         break;
 137                 }
 138                 switch (static_cast<LanguageTags>(le)) {
 139                 case LA_END: // end of structure
 140                         finished = true;
 141                         break;
 142                 case LA_AS_BABELOPTS:
 143                         lex >> as_babel_options_;
 144                         break;
 145                 case LA_BABELNAME:
 146                         lex >> babel_;
 147                         break;
 148                 case LA_POLYGLOSSIANAME:
 149                         lex >> polyglossia_name_;
 150                         break;
 151                 case LA_POLYGLOSSIAOPTS:
 152                         lex >> polyglossia_opts_;
 153                         break;
 154                 case LA_QUOTESTYLE:
 155                         lex >> quote_style_;
 156                         break;
 157                 case LA_ENCODING:
 158                         lex >> encodingStr_;
 159                         break;
 160                 case LA_FONTENC:
 161                         lex >> fontenc_;
 162                         break;
 163                 case LA_GUINAME:
 164                         lex >> display_;
 165                         break;
 166                 case LA_HAS_GUI_SUPPORT:
 167                         lex >> has_gui_support_;
 168                         break;
 169                 case LA_INTERNAL_ENC:
 170                         lex >> internal_enc_;
 171                         break;
 172                 case LA_LANG_CODE:
 173                         lex >> code_;
 174                         break;
 175                 case LA_LANG_VARIETY:
 176                         lex >> variety_;
 177                         break;
 178                 case LA_POSTBABELPREAMBLE:
 179                         babel_postsettings_ =
 180                                 lex.getLongString(from_ascii("EndPostBabelPreamble"));
 181                         break;
 182                 case LA_PREBABELPREAMBLE:
 183                         babel_presettings_ =
 184                                 lex.getLongString(from_ascii("EndPreBabelPreamble"));
 185                         break;
 186                 case LA_REQUIRES:
 187                         lex >> requires_;
 188                         break;
 189                 case LA_PROVIDES:
 190                         lex >> provides_;
 191                         break;
 192                 case LA_RTL:
 193                         lex >> rightToLeft_;
 194                         break;
 195                 }
 196         }
 197         lex.popTable();
 198         return finished && !error;
 199 }
 200
 201
 202 bool Language::read(Lexer & lex)
 203 {
 204         as_babel_options_ = 0;
 205         encoding_ = 0;
 206         internal_enc_ = 0;
 207         rightToLeft_ = 0;
 208
 209         if (!lex.next()) {
 210                 lex.printError("No name given for language: `$$Token'.");
 211                 return false;
 212         }
 213
 214         lang_ = lex.getString();
 215         LYXERR(Debug::INFO, "Reading language " << lang_);
 216         if (!readLanguage(lex)) {
 217                 LYXERR0("Error parsing language `" << lang_ << '\'');
 218                 return false;
 219         }
 220
 221         encoding_ = encodings.fromLyXName(encodingStr_);
 222         if (!encoding_ && !encodingStr_.empty()) {
 223                 encoding_ = encodings.fromLyXName("iso8859-1");
 224                 LYXERR0("Unknown encoding " << encodingStr_);
 225         }
 226         return true;
 227 }
 228
 229
 230 void Language::readLayoutTranslations(Language::TranslationMap const & trans, bool replace)
 231 {
 232         TranslationMap::const_iterator const end = trans.end();
 233         for (TranslationMap::const_iterator it = trans.begin(); it != end; ++it) {
 234                 if (replace
 235                         || layoutTranslations_.find(it->first) == layoutTranslations_.end())
 236                         layoutTranslations_[it->first] = it->second;
 237         }
 238 }
 239
 240
 241 void Languages::read(FileName const & filename)
 242 {
 243         Lexer lex;
 244         lex.setFile(filename);
 245         lex.setContext("Languages::read");
 246         while (lex.isOK()) {
 247                 int le = lex.lex();
 248                 switch (le) {
 249                 case Lexer::LEX_FEOF:
 250                         continue;
 251
 252                 default:
 253                         break;
 254                 }
 255                 if (lex.getString() != "Language") {
 256                         lex.printError("Unknown Language tag `$$Token'");
 257                         continue;
 258                 }
 259                 Language l;
 260                 l.read(lex);
 261                 if (!lex)
 262                         break;
 263                 if (l.lang() == "latex") {
 264                         // Check if latex language was not already defined.
 265                         LASSERT(latex_language == 0, continue);
 266                         static const Language latex_lang = l;
 267                         latex_language = &latex_lang;
 268                 } else if (l.lang() == "ignore") {
 269                         // Check if ignore language was not already defined.
 270                         LASSERT(ignore_language == 0, continue);
 271                         static const Language ignore_lang = l;
 272                         ignore_language = &ignore_lang;
 273                 } else
 274                         languagelist[l.lang()] = l;
 275         }
 276
 277         default_language = getLanguage("english");
 278         if (!default_language) {
 279                 LYXERR0("Default language \"english\" not found!");
 280                 default_language = &(*languagelist.begin()).second;
 281                 LYXERR0("Using \"" << default_language->lang() << "\" instead!");
 282         }
 283
 284         // Read layout translations
 285         FileName const path = libFileSearch(string(), "layouttranslations");
 286         readLayoutTranslations(path);
 287 }
 288
 289
 290 namespace {
 291
 292 bool readTranslations(Lexer & lex, Language::TranslationMap & trans)
 293 {
 294         while (lex.isOK()) {
 295                 if (lex.checkFor("End"))
 296                         break;
 297                 if (!lex.next(true))
 298                         return false;
 299                 string const key = lex.getString();
 300                 if (!lex.next(true))
 301                         return false;
 302                 docstring const val = lex.getDocString();
 303                 trans[key] = val;
 304         }
 305         return true;
 306 }
 307
 308
 309 enum Match {
 310         NoMatch,
 311         ApproximateMatch,
 312         ExactMatch
 313 };
 314
 315
 316 Match match(string const & code, Language const & lang)
 317 {
 318         // we need to mimic gettext: code can be a two-letter code, which
 319         // should match all variants, e.g. "de" should match "de_DE",
 320         // "de_AT" etc.
 321         // special case for chinese:
 322         // simplified  => code == "zh_CN", langcode == "zh_CN"
 323         // traditional => code == "zh_TW", langcode == "zh_CN"
 324         string const variety = lang.variety();
 325         string const langcode = variety.empty() ?
 326                                 lang.code() : lang.code() + '_' + variety;
 327         string const name = lang.lang();
 328         if ((code == langcode && name != "chinese-traditional")
 329                 || (code == "zh_TW"  && name == "chinese-traditional"))
 330                 return ExactMatch;
 331         if ((code.size() == 2) && (langcode.size() > 2)
 332                 && (code + '_' == langcode.substr(0, 3)))
 333                 return ApproximateMatch;
 334         return NoMatch;
 335 }
 336
 337 } // namespace
 338
 339
 340 void Languages::readLayoutTranslations(support::FileName const & filename)
 341 {
 342         Lexer lex;
 343         lex.setFile(filename);
 344         lex.setContext("Languages::read");
 345
 346         // 1) read all translations (exact and approximate matches) into trans
 347         typedef std::map<string, Language::TranslationMap> TransMap;
 348         TransMap trans;
 349         LanguageList::iterator const lbeg = languagelist.begin();
 350         LanguageList::iterator const lend = languagelist.end();
 351         while (lex.isOK()) {
 352                 if (!lex.checkFor("Translation")) {
 353                         if (lex.isOK())
 354                                 lex.printError("Unknown layout translation tag `$$Token'");
 355                         break;
 356                 }
 357                 if (!lex.next(true))
 358                         break;
 359                 string const code = lex.getString();
 360                 bool found = false;
 361                 for (LanguageList::iterator lit = lbeg; lit != lend; ++lit) {
 362                         if (match(code, lit->second) != NoMatch) {
 363                                 found = true;
 364                                 break;
 365                         }
 366                 }
 367                 if (!found) {
 368                         lex.printError("Unknown language `" + code + "'");
 369                         break;
 370                 }
 371                 if (!readTranslations(lex, trans[code])) {
 372                         lex.printError("Could not read layout translations for language `"
 373                                 + code + "'");
 374                         break;
 375                 }
 376         }
 377
 378         // 2) merge all translations into the languages
 379         // exact translations overwrite approximate ones
 380         TransMap::const_iterator const tbeg = trans.begin();
 381         TransMap::const_iterator const tend = trans.end();
 382         for (TransMap::const_iterator tit = tbeg; tit != tend; ++tit) {
 383                 for (LanguageList::iterator lit = lbeg; lit != lend; ++lit) {
 384                         Match const m = match(tit->first, lit->second);
 385                         if (m == NoMatch)
 386                                 continue;
 387                         lit->second.readLayoutTranslations(tit->second,
 388                                                            m == ExactMatch);
 389                 }
 390         }
 391
 392 }
 393
 394
 395 Language const * Languages::getLanguage(string const & language) const
 396 {
 397         if (language == "reset")
 398                 return reset_language;
 399         if (language == "ignore")
 400                 return ignore_language;
 401         const_iterator it = languagelist.find(language);
 402         return it == languagelist.end() ? reset_language : &it->second;
 403 }
 404
 405
 406 } // namespace lyx