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 #include <cstdio>
  31 #include MYTHES_H_LOCATION
  32
  33 #include "frontends/alert.h"
  34
  35 #include <algorithm>
  36 #include <cstring>
  37 #include <fstream>
  38
  39 using namespace std;
  40 using namespace lyx::support;
  41 using namespace lyx::support::os;
  42
  43 namespace lyx {
  44
  45 namespace {
  46
  47 typedef std::map<docstring, MyThes *> Thesauri;
  48
  49 } // namespace
  50
  51 struct Thesaurus::Private
  52 {
  53         ~Private()
  54         {
  55                 for (Thesauri::iterator it = thes_.begin();
  56                      it != thes_.end(); ++it) {
  57                         delete it->second;
  58                 }
  59         }
  60         ///
  61         bool thesaurusAvailable(docstring const & lang) const
  62         {
  63                 for (Thesauri::const_iterator it = thes_.begin();
  64                         it != thes_.end(); ++it) {
  65                                 if (it->first == lang)
  66                                         if (it->second)
  67                                                 return true;
  68                 }
  69                 return false;
  70         }
  71
  72         ///
  73         typedef std::pair<std::string, std::string> ThesFiles;
  74         ///
  75         ThesFiles getThesaurus(string const & path, docstring const & lang);
  76         ThesFiles getThesaurus(docstring const & lang);
  77         /// add a thesaurus to the list
  78         bool addThesaurus(docstring const & lang);
  79
  80         /// the thesauri
  81         Thesauri thes_;
  82
  83         /// the location below system/user directory
  84         /// there the data+idx files lookup will happen
  85         const string dataDirectory(void) { return "thes"; }
  86
  87 };
  88
  89
  90 pair<string,string> Thesaurus::Private::getThesaurus(string const & path, docstring const & lang)
  91 {
  92         FileName base(path);
  93         if (!base.isDirectory()) {
  94                 return make_pair(string(), string());
  95         }
  96         FileNameList const idx_files = base.dirList("idx");
  97         FileNameList const data_files = base.dirList("dat");
  98         string idx;
  99         string data;
 100         string basename;
 101
 102         LYXERR(Debug::FILES, "thesaurus path: " << path);
 103         for (FileNameList::const_iterator it = idx_files.begin(); it != idx_files.end(); ++it) {
 104                 basename = it->onlyFileNameWithoutExt();
 105                 if (contains(basename, to_ascii(lang))) {
 106                         // do not use more specific dicts.
 107                         if (contains(basename, '_') && !contains(lang, '_'))
 108                                 continue;
 109                         if (contains(basename, '-') && !contains(lang, '-'))
 110                                 continue;
 111                         ifstream ifs(it->absFileName().c_str());
 112                         if (ifs) {
 113                                 // check for appropriate version of index file
 114                                 string encoding; // first line is encoding
 115                                 int items = 0;   // second line is no. of items
 116                                 getline(ifs,encoding);
 117                                 ifs >> items;
 118                                 if (ifs.fail()) {
 119                                         LYXERR(Debug::FILES, "ignore irregular thesaurus idx file: " << it->absFileName());
 120                                         continue;
 121                                 }
 122                                 if (encoding.length() == 0 || encoding.find_first_of(',') != string::npos) {
 123                                         LYXERR(Debug::FILES, "ignore version1 thesaurus idx file: " << it->absFileName());
 124                                         continue;
 125                                 }
 126                         }
 127                         idx = it->absFileName();
 128                         LYXERR(Debug::FILES, "selected thesaurus idx file: " << idx);
 129                         break;
 130                 }
 131         }
 132         if (idx.empty()) {
 133                 // try with a more general dictionary
 134                 docstring shortcode;
 135                 if (contains(lang, '_')) {
 136                         split(lang, shortcode, '_');
 137                         LYXERR(Debug::FILES, "Did not find thesaurus for LANG code "
 138                                << lang << ". Trying with " << shortcode);
 139                         return getThesaurus(path, shortcode);
 140                 }
 141                 else if (contains(lang, '-')) {
 142                         split(lang, shortcode, '-');
 143                         LYXERR(Debug::FILES, "Did not find thesaurus for LANG code "
 144                                << lang << ". Trying with " << shortcode);
 145                         return getThesaurus(path, shortcode);
 146                 }
 147                 return make_pair(string(), string());
 148         }
 149         for (support::FileNameList::const_iterator it = data_files.begin(); it != data_files.end(); ++it) {
 150                 if (contains(it->onlyFileName(), basename)) {
 151                         data = it->absFileName();
 152                         LYXERR(Debug::FILES, "selected thesaurus data file: " << data);
 153                         break;
 154                 }
 155         }
 156         return make_pair(idx, data);
 157 }
 158
 159
 160 pair<string,string> Thesaurus::Private::getThesaurus(docstring const & lang)
 161 {
 162         string const thes_path = external_path(lyxrc.thesaurusdir_path);
 163         pair<string,string> result ;
 164
 165         if (thesaurusAvailable(lang))
 166                 return make_pair(string(), string());
 167
 168         if (!thes_path.empty())
 169                 result = getThesaurus(thes_path, lang);
 170         if (result.first.empty() || result.second.empty()) {
 171                 string const sys_path = external_path(addName(lyx::support::package().system_support().absFileName(),dataDirectory())) ;
 172                 result = getThesaurus(sys_path, lang);
 173         }
 174         if (result.first.empty() || result.second.empty()) {
 175                 string const user_path = external_path(addName(lyx::support::package().user_support().absFileName(),dataDirectory())) ;
 176                 result = getThesaurus(user_path, lang);
 177         }
 178         return result;
 179 }
 180
 181
 182 bool Thesaurus::Private::addThesaurus(docstring const & lang)
 183 {
 184         if (thesaurusAvailable(lang))
 185                 return true;
 186
 187         ThesFiles files = getThesaurus(lang);
 188         string const idx = files.first;
 189         string const data = files.second;
 190
 191         if (idx.empty() || data.empty())
 192                 return false;
 193
 194         char const * af = idx.c_str();
 195         char const * df = data.c_str();
 196         thes_[lang] = new MyThes(af, df);
 197         return true;
 198 }
 199
 200
 201 bool Thesaurus::thesaurusAvailable(docstring const & lang) const
 202 {
 203         return d->thesaurusAvailable(lang);
 204 }
 205
 206
 207 bool Thesaurus::thesaurusInstalled(docstring const & lang) const
 208 {
 209         if (thesaurusAvailable(lang))
 210                 return true;
 211         pair<string, string> files = d->getThesaurus(lang);
 212         return (!files.first.empty() && !files.second.empty());
 213 }
 214
 215
 216 Thesaurus::Meanings Thesaurus::lookup(WordLangTuple const & wl)
 217 {
 218         Meanings meanings;
 219         MyThes * mythes = 0;
 220
 221         docstring const lang_code = from_ascii(wl.lang()->code());
 222         docstring const t = wl.word();
 223
 224         if (!d->addThesaurus(lang_code))
 225                 return meanings;
 226
 227         for (Thesauri::const_iterator it = d->thes_.begin();
 228              it != d->thes_.end(); ++it) {
 229                 if (it->first == lang_code) {
 230                         mythes = it->second;
 231                         break;
 232                 }
 233         }
 234
 235         if (!mythes)
 236                 return meanings;
 237
 238         string const encoding = mythes->get_th_encoding();
 239
 240         mentry * pmean;
 241         string const text = to_iconv_encoding(support::lowercase(t), encoding);
 242         int len = strlen(text.c_str());
 243         int count = mythes->Lookup(text.c_str(), len, &pmean);
 244         if (!count) {
 245                 SpellChecker * speller = theSpellChecker();
 246                 if (!speller)
 247                         return meanings;
 248                 docstring_list suggestions;
 249                 speller->stem(wl, suggestions);
 250                 for (size_t i = 0; i != suggestions.size(); ++i) {
 251                         string const wordform = to_iconv_encoding(support::lowercase(suggestions[i]), encoding);
 252                         len = strlen(wordform.c_str());
 253                         count = mythes->Lookup(wordform.c_str(), len, &pmean);
 254                         if (count)
 255                                 break;
 256                 }
 257                 if (!count)
 258                         return meanings;
 259         }
 260
 261         // don't change value of pmean or count
 262         // they are needed for the CleanUpAfterLookup routine
 263         mentry * pm = pmean;
 264         docstring meaning;
 265         for (int i = 0; i < count; i++) {
 266                 vector<docstring> ret;
 267                 meaning = from_iconv_encoding(string(pm->defn), encoding);
 268                 // remove silly item
 269                 if (support::prefixIs(meaning, '-'))
 270                         meaning = support::ltrim(meaning, "- ");
 271                 for (int j = 0; j < pm->count; j++) {
 272                         ret.push_back(from_iconv_encoding(string(pm->psyns[j]), encoding));
 273                 }
 274                 meanings[meaning] = ret;
 275                 ++pm;
 276         }
 277         // now clean up all allocated memory
 278         mythes->CleanUpAfterLookup(&pmean, count);
 279
 280         for (Meanings::iterator it = meanings.begin();
 281              it != meanings.end(); ++it)
 282                 sort(it->second.begin(), it->second.end());
 283
 284         return meanings;
 285 }
 286
 287
 288 Thesaurus::Thesaurus()
 289         : d(new Thesaurus::Private)
 290 {
 291 }
 292
 293
 294 Thesaurus::~Thesaurus()
 295 {
 296         delete d;
 297 }
 298
 299 // Global instance
 300 Thesaurus thesaurus;
 301
 302
 303 } // namespace lyx