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                         // do not use more specific dicts.
 111                         if (contains(basename, '_') && !contains(lang, '_'))
 112                                 continue;
 113                         if (contains(basename, '-') && !contains(lang, '-'))
 114                                 continue;
 115                         ifstream ifs(it->absFileName().c_str());
 116                         if (ifs) {
 117                                 // check for appropriate version of index file
 118                                 string encoding; // first line is encoding
 119                                 int items = 0;   // second line is no. of items
 120                                 getline(ifs,encoding);
 121                                 ifs >> items;
 122                                 if (ifs.fail()) {
 123                                         LYXERR(Debug::FILES, "ignore irregular thesaurus idx file: " << it->absFileName());
 124                                         continue;
 125                                 }
 126                                 if (encoding.length() == 0 || encoding.find_first_of(',') != string::npos) {
 127                                         LYXERR(Debug::FILES, "ignore version1 thesaurus idx file: " << it->absFileName());
 128                                         continue;
 129                                 }
 130                         }
 131                         idx = it->absFileName();
 132                         LYXERR(Debug::FILES, "selected thesaurus idx file: " << idx);
 133                         break;
 134                 }
 135         }
 136         if (idx.empty()) {
 137                 // try with a more general dictionary
 138                 docstring shortcode;
 139                 if (contains(lang, '_')) {
 140                         split(lang, shortcode, '_');
 141                         LYXERR(Debug::FILES, "Did not find thesaurus for LANG code "
 142                                << lang << ". Trying with " << shortcode);
 143                         return getThesaurus(path, shortcode);
 144                 }
 145                 else if (contains(lang, '-')) {
 146                         split(lang, shortcode, '-');
 147                         LYXERR(Debug::FILES, "Did not find thesaurus for LANG code "
 148                                << lang << ". Trying with " << shortcode);
 149                         return getThesaurus(path, shortcode);
 150                 }
 151                 return make_pair(string(), string());
 152         }
 153         for (support::FileNameList::const_iterator it = data_files.begin(); it != data_files.end(); ++it) {
 154                 if (contains(it->onlyFileName(), basename)) {
 155                         data = it->absFileName();
 156                         LYXERR(Debug::FILES, "selected thesaurus data file: " << data);
 157                         break;
 158                 }
 159         }
 160         return make_pair(idx, data);
 161 }
 162
 163
 164 pair<string,string> Thesaurus::Private::getThesaurus(docstring const & lang)
 165 {
 166         string const thes_path = external_path(lyxrc.thesaurusdir_path);
 167         pair<string,string> result ;
 168
 169         if (thesaurusAvailable(lang))
 170                 return make_pair(string(), string());
 171
 172         if (!thes_path.empty())
 173                 result = getThesaurus(thes_path, lang);
 174         if (result.first.empty() || result.second.empty()) {
 175                 string const sys_path = external_path(addName(lyx::support::package().system_support().absFileName(),dataDirectory())) ;
 176                 result = getThesaurus(sys_path, lang);
 177         }
 178         if (result.first.empty() || result.second.empty()) {
 179                 string const user_path = external_path(addName(lyx::support::package().user_support().absFileName(),dataDirectory())) ;
 180                 result = getThesaurus(user_path, lang);
 181         }
 182         return result;
 183 }
 184
 185
 186 bool Thesaurus::Private::addThesaurus(docstring const & lang)
 187 {
 188         if (thesaurusAvailable(lang))
 189                 return true;
 190
 191         ThesFiles files = getThesaurus(lang);
 192         string const idx = files.first;
 193         string const data = files.second;
 194
 195         if (idx.empty() || data.empty())
 196                 return false;
 197
 198         char const * af = idx.c_str();
 199         char const * df = data.c_str();
 200         thes_[lang] = new MyThes(af, df);
 201         return true;
 202 }
 203
 204
 205 bool Thesaurus::thesaurusAvailable(docstring const & lang) const
 206 {
 207         return d->thesaurusAvailable(lang);
 208 }
 209
 210
 211 bool Thesaurus::thesaurusInstalled(docstring const & lang) const
 212 {
 213         if (thesaurusAvailable(lang))
 214                 return true;
 215         pair<string, string> files = d->getThesaurus(lang);
 216         return (!files.first.empty() && !files.second.empty());
 217 }
 218
 219
 220 Thesaurus::Meanings Thesaurus::lookup(WordLangTuple const & wl)
 221 {
 222         Meanings meanings;
 223         MyThes * mythes = 0;
 224
 225         docstring const lang_code = from_ascii(wl.lang()->code());
 226         docstring const t = wl.word();
 227
 228         if (!d->addThesaurus(lang_code))
 229                 return meanings;
 230
 231         for (Thesauri::const_iterator it = d->thes_.begin();
 232              it != d->thes_.end(); ++it) {
 233                 if (it->first == lang_code) {
 234                         mythes = it->second;
 235                         break;
 236                 }
 237         }
 238
 239         if (!mythes)
 240                 return meanings;
 241
 242         string const encoding = mythes->get_th_encoding();
 243
 244         mentry * pmean;
 245         string const text = to_iconv_encoding(support::lowercase(t), encoding);
 246         int len = strlen(text.c_str());
 247         int count = mythes->Lookup(text.c_str(), len, &pmean);
 248         if (!count) {
 249                 SpellChecker * speller = theSpellChecker();
 250                 if (!speller)
 251                         return meanings;
 252                 docstring_list suggestions;
 253                 speller->stem(wl, suggestions);
 254                 for (size_t i = 0; i != suggestions.size(); ++i) {
 255                         string const wordform = to_iconv_encoding(support::lowercase(suggestions[i]), encoding);
 256                         len = strlen(wordform.c_str());
 257                         count = mythes->Lookup(wordform.c_str(), len, &pmean);
 258                         if (count)
 259                                 break;
 260                 }
 261                 if (!count)
 262                         return meanings;
 263         }
 264
 265         // don't change value of pmean or count
 266         // they are needed for the CleanUpAfterLookup routine
 267         mentry * pm = pmean;
 268         docstring meaning;
 269         for (int i = 0; i < count; i++) {
 270                 vector<docstring> ret;
 271                 meaning = from_iconv_encoding(string(pm->defn), encoding);
 272                 // remove silly item
 273                 if (support::prefixIs(meaning, '-'))
 274                         meaning = support::ltrim(meaning, "- ");
 275                 for (int j = 0; j < pm->count; j++) {
 276                         ret.push_back(from_iconv_encoding(string(pm->psyns[j]), encoding));
 277                 }
 278                 meanings[meaning] = ret;
 279                 ++pm;
 280         }
 281         // now clean up all allocated memory
 282         mythes->CleanUpAfterLookup(&pmean, count);
 283
 284         for (Meanings::iterator it = meanings.begin();
 285              it != meanings.end(); ++it)
 286                 sort(it->second.begin(), it->second.end());
 287
 288         return meanings;
 289 }
 290
 291
 292 Thesaurus::Thesaurus()
 293         : d(new Thesaurus::Private)
 294 {
 295 }
 296
 297
 298 Thesaurus::~Thesaurus()
 299 {
 300         delete d;
 301 }
 302
 303 // Global instance
 304 Thesaurus thesaurus;
 305
 306
 307 } // namespace lyx