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