src/Thesaurus.cpp

   1 /**
   2  * \file Thesaurus.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author John Levon
   7  * \author Jürgen Spitzmüller
   8  *
   9  * Full author contact details are available in file CREDITS.
  10  */
  11
  12 #include <config.h>
  13
  14 #include "Thesaurus.h"
  15
  16 #include "support/debug.h"
  17 #include "support/gettext.h"
  18 #include "LyXRC.h"
  19
  20 #include "support/FileNameList.h"
  21 #include "support/filetools.h"
  22 #include "support/lstrings.h"
  23 #include "support/os.h"
  24 #include "support/unicode.h"
  25
  26 #include "frontends/alert.h"
  27
  28 #include <algorithm>
  29 #include <cstring>
  30
  31 using namespace std;
  32 using namespace lyx::support;
  33 using namespace lyx::support::os;
  34
  35 namespace lyx {
  36
  37 #ifndef HAVE_LIBMYTHES
  38 #ifdef HAVE_LIBAIKSAURUS
  39
  40
  41 Thesaurus::Thesaurus()
  42         : thes_(new Aiksaurus)
  43 {}
  44
  45
  46 Thesaurus::~Thesaurus()
  47 {
  48         delete thes_;
  49 }
  50
  51
  52 Thesaurus::Meanings Thesaurus::lookup(docstring const & t, docstring const &)
  53 {
  54         Meanings meanings;
  55
  56         // aiksaurus is for english text only, therefore it does not work
  57         // with non-ascii strings.
  58         // The interface of the Thesaurus class uses docstring because a
  59         // non-english thesaurus is possible in theory.
  60         if (!support::isAscii(t))
  61                 // to_ascii() would assert
  62                 return meanings;
  63
  64         string const text = to_ascii(t);
  65
  66         docstring error = from_ascii(thes_->error());
  67         if (!error.empty()) {
  68                 static bool sent_error = false;
  69                 if (!sent_error) {
  70                         frontend::Alert::error(_("Thesaurus failure"),
  71                                      bformat(_("Aiksaurus returned the following error:\n\n%1$s."),
  72                                              error));
  73                         sent_error = true;
  74                 }
  75                 return meanings;
  76         }
  77         if (!thes_->find(text.c_str()))
  78                 return meanings;
  79
  80         // weird api, but ...
  81
  82         int prev_meaning = -1;
  83         int cur_meaning;
  84         docstring meaning;
  85
  86         // correct, returns "" at the end
  87         string ret = thes_->next(cur_meaning);
  88
  89         while (!ret.empty()) {
  90                 if (cur_meaning != prev_meaning) {
  91                         meaning = from_ascii(ret);
  92                         ret = thes_->next(cur_meaning);
  93                         prev_meaning = cur_meaning;
  94                 } else {
  95                         if (ret != text)
  96                                 meanings[meaning].push_back(from_ascii(ret));
  97                 }
  98
  99                 ret = thes_->next(cur_meaning);
 100         }
 101
 102         for (Meanings::iterator it = meanings.begin();
 103              it != meanings.end(); ++it)
 104                 sort(it->second.begin(), it->second.end());
 105
 106         return meanings;
 107 }
 108
 109
 110 bool Thesaurus::thesaurusAvailable(docstring const & lang) const
 111 {
 112         // we support English only
 113         return prefixIs(lang, from_ascii("en_"));
 114 }
 115
 116 #endif // HAVE_LIBAIKSAURUS
 117 #endif // !HAVE_LIBMYTHES
 118
 119
 120 #ifdef HAVE_LIBMYTHES
 121
 122 namespace {
 123
 124 string const to_iconv_encoding(docstring const & s, string const & encoding)
 125 {
 126         std::vector<char> const encoded =
 127                 ucs4_to_eightbit(s.data(), s.length(), encoding);
 128         return string(encoded.begin(), encoded.end());
 129 }
 130
 131
 132 docstring const from_iconv_encoding(string const & s, string const & encoding)
 133 {
 134         std::vector<char_type> const ucs4 =
 135                 eightbit_to_ucs4(s.data(), s.length(), encoding);
 136         return docstring(ucs4.begin(), ucs4.end());
 137 }
 138
 139 } // namespace anon
 140
 141
 142 Thesaurus::Thesaurus()
 143 {}
 144
 145
 146 Thesaurus::~Thesaurus()
 147 {
 148         for (Thesauri::iterator it = thes_.begin();
 149              it != thes_.end(); ++it) {
 150                 delete it->second;
 151         }
 152 }
 153
 154
 155 bool Thesaurus::addThesaurus(docstring const & lang)
 156 {
 157         string const thes_path = external_path(lyxrc.thesaurusdir_path);
 158         LYXERR(Debug::FILES, "thesaurus path: " << thes_path);
 159         if (thes_path.empty())
 160                 return false;
 161
 162         if (thesaurusAvailable(lang))
 163                 return true;
 164
 165         FileNameList const idx_files = FileName(thes_path).dirList("idx");
 166         FileNameList const data_files = FileName(thes_path).dirList("dat");
 167         string idx;
 168         string data;
 169
 170         for (FileNameList::const_iterator it = idx_files.begin();
 171              it != idx_files.end(); ++it) {
 172                 LYXERR(Debug::FILES, "found thesaurus idx file: " << it->onlyFileName());
 173                 if (contains(it->onlyFileName(), to_ascii(lang))) {
 174                         idx = it->absFilename();
 175                         LYXERR(Debug::FILES, "selected thesaurus idx file: " << idx);
 176                         break;
 177                         }
 178                 }
 179
 180         for (support::FileNameList::const_iterator it = data_files.begin();
 181              it != data_files.end(); ++it) {
 182                 LYXERR(Debug::FILES, "found thesaurus data file: " << it->onlyFileName());
 183                 if (contains(it->onlyFileName(), to_ascii(lang))) {
 184                         data = it->absFilename();
 185                         LYXERR(Debug::FILES, "selected thesaurus data file: " << data);
 186                         break;
 187                         }
 188                 }
 189
 190         if (idx.empty() || data.empty())
 191                 return false;
 192
 193         char const * af = idx.c_str();
 194         char const * df = data.c_str();
 195         thes_[lang] = new MyThes(af, df);
 196         return true;
 197 }
 198
 199
 200 bool Thesaurus::thesaurusAvailable(docstring const & lang) const
 201 {
 202         for (Thesauri::const_iterator it = thes_.begin();
 203              it != thes_.end(); ++it) {
 204                 if (it->first == lang)
 205                         if (it->second)
 206                                 return true;
 207         }
 208
 209         return false;
 210 }
 211
 212
 213 Thesaurus::Meanings Thesaurus::lookup(docstring const & t, docstring const & lang)
 214 {
 215         Meanings meanings;
 216         MyThes * mythes = 0;
 217
 218         if (!addThesaurus(lang))
 219                 return meanings;
 220
 221         for (Thesauri::const_iterator it = thes_.begin();
 222              it != thes_.end(); ++it) {
 223                 if (it->first == lang) {
 224                         mythes = it->second;
 225                         break;
 226                 }
 227         }
 228
 229         if (!mythes)
 230                 return meanings;
 231
 232         string const encoding = mythes->get_th_encoding();
 233
 234         mentry * pmean;
 235         string const text = to_iconv_encoding(support::lowercase(t), encoding);
 236         int len = strlen(text.c_str());
 237         int count = mythes->Lookup(text.c_str(), len, &pmean);
 238         if (!count)
 239                 return meanings;
 240
 241         // don't change value of pmean or count
 242         // they are needed for the CleanUpAfterLookup routine
 243         mentry * pm = pmean;
 244         docstring meaning;
 245         docstring ret;
 246         for (int i = 0; i < count; i++) {
 247                 meaning = from_iconv_encoding(string(pm->defn), encoding);
 248                 // remove silly item
 249                 if (support::prefixIs(meaning, '-'))
 250                         meaning = support::ltrim(meaning, "- ");
 251                 for (int j = 0; j < pm->count; j++) {
 252                         ret = from_iconv_encoding(string(pm->psyns[j]), encoding);
 253                 }
 254         meanings[meaning].push_back(ret);
 255         pm++;
 256         }
 257         // now clean up all allocated memory
 258         mythes->CleanUpAfterLookup(&pmean, count);
 259
 260         for (Meanings::iterator it = meanings.begin();
 261              it != meanings.end(); ++it)
 262                 sort(it->second.begin(), it->second.end());
 263
 264         return meanings;
 265 }
 266
 267 #else
 268 #ifndef HAVE_LIBAIKSAURUS
 269 Thesaurus::Thesaurus()
 270 {
 271 }
 272
 273
 274 Thesaurus::~Thesaurus()
 275 {
 276 }
 277
 278
 279 Thesaurus::Meanings Thesaurus::lookup(docstring const &, docstring const &)
 280 {
 281         return Meanings();
 282 }
 283
 284
 285 bool Thesaurus::thesaurusAvailable(docstring const & lang) const
 286 {
 287         return false;
 288 }
 289
 290 #endif
 291 #endif // HAVE_LIBMYTHES
 292
 293 // Global instance
 294 Thesaurus thesaurus;
 295
 296
 297 } // namespace lyx