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 "LyXRC.h"
  17
  18 #include "support/FileNameList.h"
  19 #include "support/debug.h"
  20 #include "support/filetools.h"
  21 #include "support/gettext.h"
  22 #include "support/lstrings.h"
  23 #include "support/os.h"
  24 #include "support/unicode.h"
  25
  26 #include "support/mythes/mythes.hxx"
  27
  28 #include "frontends/alert.h"
  29
  30 #include <algorithm>
  31 #include <cstring>
  32
  33 using namespace std;
  34 using namespace lyx::support;
  35 using namespace lyx::support::os;
  36
  37 namespace lyx {
  38
  39 namespace {
  40
  41 string const to_iconv_encoding(docstring const & s, string const & encoding)
  42 {
  43         std::vector<char> const encoded =
  44                 ucs4_to_eightbit(s.data(), s.length(), encoding);
  45         return string(encoded.begin(), encoded.end());
  46 }
  47
  48
  49 docstring const from_iconv_encoding(string const & s, string const & encoding)
  50 {
  51         std::vector<char_type> const ucs4 =
  52                 eightbit_to_ucs4(s.data(), s.length(), encoding);
  53         return docstring(ucs4.begin(), ucs4.end());
  54 }
  55
  56 typedef std::map<docstring, MyThes *> Thesauri;
  57
  58 } // namespace anon
  59
  60
  61 struct Thesaurus::Private
  62 {
  63         ~Private()
  64         {
  65                 for (Thesauri::iterator it = thes_.begin();
  66                      it != thes_.end(); ++it) {
  67                         delete it->second;
  68                 }
  69         }
  70         ///
  71         bool thesaurusAvailable(docstring const & lang) const
  72         {
  73                 for (Thesauri::const_iterator it = thes_.begin();
  74                         it != thes_.end(); ++it) {
  75                                 if (it->first == lang)
  76                                         if (it->second)
  77                                                 return true;
  78                 }
  79                 return false;
  80         }
  81
  82         /// add a thesaurus to the list
  83         bool addThesaurus(docstring const & lang);
  84
  85         /// the thesauri
  86         Thesauri thes_;
  87 };
  88
  89 bool Thesaurus::Private::addThesaurus(docstring const & lang)
  90 {
  91         string const thes_path = external_path(lyxrc.thesaurusdir_path);
  92         LYXERR(Debug::FILES, "thesaurus path: " << thes_path);
  93         if (thes_path.empty())
  94                 return false;
  95
  96         if (thesaurusAvailable(lang))
  97                 return true;
  98
  99         FileNameList const idx_files = FileName(thes_path).dirList("idx");
 100         FileNameList const data_files = FileName(thes_path).dirList("dat");
 101         string idx;
 102         string data;
 103
 104         for (FileNameList::const_iterator it = idx_files.begin();
 105              it != idx_files.end(); ++it) {
 106                 LYXERR(Debug::FILES, "found thesaurus idx file: " << it->onlyFileName());
 107                 if (contains(it->onlyFileName(), to_ascii(lang))) {
 108                         idx = it->absFilename();
 109                         LYXERR(Debug::FILES, "selected thesaurus idx file: " << idx);
 110                         break;
 111                         }
 112                 }
 113
 114         for (support::FileNameList::const_iterator it = data_files.begin();
 115              it != data_files.end(); ++it) {
 116                 LYXERR(Debug::FILES, "found thesaurus data file: " << it->onlyFileName());
 117                 if (contains(it->onlyFileName(), to_ascii(lang))) {
 118                         data = it->absFilename();
 119                         LYXERR(Debug::FILES, "selected thesaurus data file: " << data);
 120                         break;
 121                         }
 122                 }
 123
 124         if (idx.empty() || data.empty())
 125                 return false;
 126
 127         char const * af = idx.c_str();
 128         char const * df = data.c_str();
 129         thes_[lang] = new MyThes(af, df);
 130         return true;
 131 }
 132
 133
 134 bool Thesaurus::thesaurusAvailable(docstring const & lang) const
 135 {
 136         return d->thesaurusAvailable(lang);
 137 }
 138
 139
 140 Thesaurus::Meanings Thesaurus::lookup(docstring const & t, docstring const & lang)
 141 {
 142         Meanings meanings;
 143         MyThes * mythes = 0;
 144
 145         if (!d->addThesaurus(lang))
 146                 return meanings;
 147
 148         for (Thesauri::const_iterator it = d->thes_.begin();
 149              it != d->thes_.end(); ++it) {
 150                 if (it->first == lang) {
 151                         mythes = it->second;
 152                         break;
 153                 }
 154         }
 155
 156         if (!mythes)
 157                 return meanings;
 158
 159         string const encoding = mythes->get_th_encoding();
 160
 161         mentry * pmean;
 162         string const text = to_iconv_encoding(support::lowercase(t), encoding);
 163         int len = strlen(text.c_str());
 164         int count = mythes->Lookup(text.c_str(), len, &pmean);
 165         if (!count)
 166                 return meanings;
 167
 168         // don't change value of pmean or count
 169         // they are needed for the CleanUpAfterLookup routine
 170         mentry * pm = pmean;
 171         docstring meaning;
 172         docstring ret;
 173         for (int i = 0; i < count; i++) {
 174                 meaning = from_iconv_encoding(string(pm->defn), encoding);
 175                 // remove silly item
 176                 if (support::prefixIs(meaning, '-'))
 177                         meaning = support::ltrim(meaning, "- ");
 178                 for (int j = 0; j < pm->count; j++) {
 179                         ret = from_iconv_encoding(string(pm->psyns[j]), encoding);
 180                 }
 181         meanings[meaning].push_back(ret);
 182         pm++;
 183         }
 184         // now clean up all allocated memory
 185         mythes->CleanUpAfterLookup(&pmean, count);
 186
 187         for (Meanings::iterator it = meanings.begin();
 188              it != meanings.end(); ++it)
 189                 sort(it->second.begin(), it->second.end());
 190
 191         return meanings;
 192 }
 193
 194
 195 Thesaurus::Thesaurus() : d(new Thesaurus::Private)
 196 {
 197 }
 198
 199
 200 Thesaurus::~Thesaurus()
 201 {
 202         delete d;
 203 }
 204
 205 // Global instance
 206 Thesaurus thesaurus;
 207
 208
 209 } // namespace lyx