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