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