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 "SpellChecker.h"
  19 #include "WordLangTuple.h"
  20
  21 #include "support/FileNameList.h"
  22 #include "support/Package.h"
  23 #include "support/debug.h"
  24 #include "support/docstring_list.h"
  25 #include "support/filetools.h"
  26 #include "support/gettext.h"
  27 #include "support/lstrings.h"
  28 #include "support/os.h"
  29
  30 #ifdef USE_EXTERNAL_MYTHES
  31 #include MYTHES_H_LOCATION
  32 #else
  33 #include <cstdio>
  34 #include "support/mythes/mythes.hxx"
  35 #endif
  36
  37 #include "frontends/alert.h"
  38
  39 #include <algorithm>
  40 #include <cstring>
  41 #include <fstream>
  42
  43 using namespace std;
  44 using namespace lyx::support;
  45 using namespace lyx::support::os;
  46
  47 namespace lyx {
  48
  49 namespace {
  50
  51 typedef std::map<docstring, MyThes *> Thesauri;
  52
  53 } // namespace anon
  54
  55 struct Thesaurus::Private
  56 {
  57         ~Private()
  58         {
  59                 for (Thesauri::iterator it = thes_.begin();
  60                      it != thes_.end(); ++it) {
  61                         delete it->second;
  62                 }
  63         }
  64         ///
  65         bool thesaurusAvailable(docstring const & lang) const
  66         {
  67                 for (Thesauri::const_iterator it = thes_.begin();
  68                         it != thes_.end(); ++it) {
  69                                 if (it->first == lang)
  70                                         if (it->second)
  71                                                 return true;
  72                 }
  73                 return false;
  74         }
  75
  76         ///
  77         typedef std::pair<std::string, std::string> ThesFiles;
  78         ///
  79         ThesFiles getThesaurus(string const & path, docstring const & lang);
  80         ThesFiles getThesaurus(docstring const & lang);
  81         /// add a thesaurus to the list
  82         bool addThesaurus(docstring const & lang);
  83
  84         /// the thesauri
  85         Thesauri thes_;
  86
  87         /// the location below system/user directory
  88         /// there the data+idx files lookup will happen
  89         const string dataDirectory(void) { return "thes"; }
  90
  91 };
  92
  93
  94 pair<string,string> Thesaurus::Private::getThesaurus(string const & path, docstring const & lang)
  95 {
  96         FileName base(path);
  97         if (!base.isDirectory()) {
  98                 return make_pair(string(), string());
  99         }
 100         FileNameList const idx_files = base.dirList("idx");
 101         FileNameList const data_files = base.dirList("dat");
 102         string idx;
 103         string data;
 104         string basename;
 105
 106         LYXERR(Debug::FILES, "thesaurus path: " << path);
 107         for (FileNameList::const_iterator it = idx_files.begin(); it != idx_files.end(); ++it) {
 108                 basename = it->onlyFileNameWithoutExt();
 109                 if (contains(basename, to_ascii(lang))) {
 110                         ifstream ifs(it->absFileName().c_str());
 111                         if (ifs) {
 112                                 // check for appropriate version of index file
 113                                 string encoding; // first line is encoding
 114                                 int items = 0;   // second line is no. of items
 115                                 getline(ifs,encoding);
 116                                 ifs >> items;
 117                                 if (ifs.fail()) {
 118                                         LYXERR(Debug::FILES, "ignore irregular thesaurus idx file: " << it->absFileName());
 119                                         continue;
 120                                 }
 121                                 if (encoding.length() == 0 || encoding.find_first_of(',') != string::npos) {
 122                                         LYXERR(Debug::FILES, "ignore version1 thesaurus idx file: " << it->absFileName());
 123                                         continue;
 124                                 }
 125                         }
 126                         idx = it->absFileName();
 127                         LYXERR(Debug::FILES, "selected thesaurus idx file: " << idx);
 128                         break;
 129                 }
 130         }
 131         if (idx.empty())
 132                 return make_pair(string(), string());
 133         for (support::FileNameList::const_iterator it = data_files.begin(); it != data_files.end(); ++it) {
 134                 if (contains(it->onlyFileName(), basename)) {
 135                         data = it->absFileName();
 136                         LYXERR(Debug::FILES, "selected thesaurus data file: " << data);
 137                         break;
 138                 }
 139         }
 140         return make_pair(idx, data);
 141 }
 142
 143
 144 pair<string,string> Thesaurus::Private::getThesaurus(docstring const & lang)
 145 {
 146         string const thes_path = external_path(lyxrc.thesaurusdir_path);
 147         pair<string,string> result ;
 148
 149         if (thesaurusAvailable(lang))
 150                 return make_pair(string(), string());
 151
 152         if (!thes_path.empty())
 153                 result = getThesaurus(thes_path, lang);
 154         if (result.first.empty() || result.second.empty()) {
 155                 string const sys_path = external_path(addName(lyx::support::package().system_support().absFileName(),dataDirectory())) ;
 156                 result = getThesaurus(sys_path, lang);
 157         }
 158         if (result.first.empty() || result.second.empty()) {
 159                 string const user_path = external_path(addName(lyx::support::package().user_support().absFileName(),dataDirectory())) ;
 160                 result = getThesaurus(user_path, lang);
 161         }
 162         return result;
 163 }
 164
 165
 166 bool Thesaurus::Private::addThesaurus(docstring const & lang)
 167 {
 168         if (thesaurusAvailable(lang))
 169                 return true;
 170
 171         ThesFiles files = getThesaurus(lang);
 172         string const idx = files.first;
 173         string const data = files.second;
 174
 175         if (idx.empty() || data.empty())
 176                 return false;
 177
 178         char const * af = idx.c_str();
 179         char const * df = data.c_str();
 180         thes_[lang] = new MyThes(af, df);
 181         return true;
 182 }
 183
 184
 185 bool Thesaurus::thesaurusAvailable(docstring const & lang) const
 186 {
 187         return d->thesaurusAvailable(lang);
 188 }
 189
 190
 191 bool Thesaurus::thesaurusInstalled(docstring const & lang) const
 192 {
 193         if (thesaurusAvailable(lang))
 194                 return true;
 195         pair<string, string> files = d->getThesaurus(lang);
 196         return (!files.first.empty() && !files.second.empty());
 197 }
 198
 199
 200 Thesaurus::Meanings Thesaurus::lookup(WordLangTuple const & wl)
 201 {
 202         Meanings meanings;
 203         MyThes * mythes = 0;
 204
 205         docstring const lang_code = from_ascii(wl.lang()->code());
 206         docstring const t = wl.word();
 207
 208         if (!d->addThesaurus(lang_code))
 209                 return meanings;
 210
 211         for (Thesauri::const_iterator it = d->thes_.begin();
 212              it != d->thes_.end(); ++it) {
 213                 if (it->first == lang_code) {
 214                         mythes = it->second;
 215                         break;
 216                 }
 217         }
 218
 219         if (!mythes)
 220                 return meanings;
 221
 222         string const encoding = mythes->get_th_encoding();
 223
 224         mentry * pmean;
 225         string const text = to_iconv_encoding(support::lowercase(t), encoding);
 226         int len = strlen(text.c_str());
 227         int count = mythes->Lookup(text.c_str(), len, &pmean);
 228         if (!count) {
 229                 SpellChecker * speller = theSpellChecker();
 230                 if (!speller)
 231                         return meanings;
 232                 docstring_list suggestions;
 233                 speller->stem(wl, suggestions);
 234                 for (size_t i = 0; i != suggestions.size(); ++i) {
 235                         string const wordform = to_iconv_encoding(support::lowercase(suggestions[i]), encoding);
 236                         len = strlen(wordform.c_str());
 237                         count = mythes->Lookup(wordform.c_str(), len, &pmean);
 238                         if (count)
 239                                 break;
 240                 }
 241                 if (!count)
 242                         return meanings;
 243         }
 244
 245         // don't change value of pmean or count
 246         // they are needed for the CleanUpAfterLookup routine
 247         mentry * pm = pmean;
 248         docstring meaning;
 249         for (int i = 0; i < count; i++) {
 250                 vector<docstring> ret;
 251                 meaning = from_iconv_encoding(string(pm->defn), encoding);
 252                 // remove silly item
 253                 if (support::prefixIs(meaning, '-'))
 254                         meaning = support::ltrim(meaning, "- ");
 255                 for (int j = 0; j < pm->count; j++) {
 256                         ret.push_back(from_iconv_encoding(string(pm->psyns[j]), encoding));
 257                 }
 258                 meanings[meaning] = ret;
 259                 ++pm;
 260         }
 261         // now clean up all allocated memory
 262         mythes->CleanUpAfterLookup(&pmean, count);
 263
 264         for (Meanings::iterator it = meanings.begin();
 265              it != meanings.end(); ++it)
 266                 sort(it->second.begin(), it->second.end());
 267
 268         return meanings;
 269 }
 270
 271
 272 Thesaurus::Thesaurus()
 273         : d(new Thesaurus::Private)
 274 {
 275 }
 276
 277
 278 Thesaurus::~Thesaurus()
 279 {
 280         delete d;
 281 }
 282
 283 // Global instance
 284 Thesaurus thesaurus;
 285
 286
 287 } // namespace lyx