3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
7 * \author Jürgen Spitzmüller
9 * Full author contact details are available in file CREDITS.
14 #include "Thesaurus.h"
16 #include "support/debug.h"
17 #include "support/gettext.h"
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"
26 #include "frontends/alert.h"
29 #include MYTHES_H_LOCATION
31 #ifdef HAVE_LIBAIKSAURUS
32 #include AIKSAURUS_H_LOCATION
33 #endif // HAVE_LIBAIKSAURUS
34 #endif // !HAVE_LIBMYTHES
40 using namespace lyx::support;
41 using namespace lyx::support::os;
45 #ifdef HAVE_LIBAIKSAURUS
47 struct Thesaurus::Private
49 Private(): thes_(new Aiksaurus) {}
53 Thesaurus::Meanings Thesaurus::lookup(docstring const & t, docstring const &)
57 // aiksaurus is for english text only, therefore it does not work
58 // with non-ascii strings.
59 // The interface of the Thesaurus class uses docstring because a
60 // non-english thesaurus is possible in theory.
61 if (!support::isAscii(t))
62 // to_ascii() would assert
65 string const text = to_ascii(t);
67 docstring error = from_ascii(d->thes_->error());
69 static bool sent_error = false;
71 frontend::Alert::error(_("Thesaurus failure"),
72 bformat(_("Aiksaurus returned the following error:\n\n%1$s."),
78 if (!d->thes_->find(text.c_str()))
83 int prev_meaning = -1;
87 // correct, returns "" at the end
88 string ret = d->thes_->next(cur_meaning);
90 while (!ret.empty()) {
91 if (cur_meaning != prev_meaning) {
92 meaning = from_ascii(ret);
93 ret = d->thes_->next(cur_meaning);
94 prev_meaning = cur_meaning;
97 meanings[meaning].push_back(from_ascii(ret));
100 ret = d->thes_->next(cur_meaning);
103 for (Meanings::iterator it = meanings.begin();
104 it != meanings.end(); ++it)
105 sort(it->second.begin(), it->second.end());
111 bool Thesaurus::thesaurusAvailable(docstring const & lang) const
113 // we support English only
114 return prefixIs(lang, from_ascii("en_"));
117 #else // HAVE_LIBAIKSAURUS
118 #ifdef HAVE_LIBMYTHES
122 string const to_iconv_encoding(docstring const & s, string const & encoding)
124 std::vector<char> const encoded =
125 ucs4_to_eightbit(s.data(), s.length(), encoding);
126 return string(encoded.begin(), encoded.end());
130 docstring const from_iconv_encoding(string const & s, string const & encoding)
132 std::vector<char_type> const ucs4 =
133 eightbit_to_ucs4(s.data(), s.length(), encoding);
134 return docstring(ucs4.begin(), ucs4.end());
137 typedef std::map<docstring, MyThes *> Thesauri;
142 struct Thesaurus::Private
146 for (Thesauri::iterator it = thes_.begin();
147 it != thes_.end(); ++it) {
152 /// add a thesaurus to the list
153 bool addThesaurus(docstring const & lang);
159 bool Thesaurus::Private::addThesaurus(docstring const & lang)
161 string const thes_path = external_path(lyxrc.thesaurusdir_path);
162 LYXERR(Debug::FILES, "thesaurus path: " << thes_path);
163 if (thes_path.empty())
166 if (thesaurusAvailable(lang))
169 FileNameList const idx_files = FileName(thes_path).dirList("idx");
170 FileNameList const data_files = FileName(thes_path).dirList("dat");
174 for (FileNameList::const_iterator it = idx_files.begin();
175 it != idx_files.end(); ++it) {
176 LYXERR(Debug::FILES, "found thesaurus idx file: " << it->onlyFileName());
177 if (contains(it->onlyFileName(), to_ascii(lang))) {
178 idx = it->absFilename();
179 LYXERR(Debug::FILES, "selected thesaurus idx file: " << idx);
184 for (support::FileNameList::const_iterator it = data_files.begin();
185 it != data_files.end(); ++it) {
186 LYXERR(Debug::FILES, "found thesaurus data file: " << it->onlyFileName());
187 if (contains(it->onlyFileName(), to_ascii(lang))) {
188 data = it->absFilename();
189 LYXERR(Debug::FILES, "selected thesaurus data file: " << data);
194 if (idx.empty() || data.empty())
197 char const * af = idx.c_str();
198 char const * df = data.c_str();
199 thes_[lang] = new MyThes(af, df);
204 bool Thesaurus::thesaurusAvailable(docstring const & lang) const
206 for (Thesauri::const_iterator it = d->thes_.begin();
207 it != d->thes_.end(); ++it) {
208 if (it->first == lang)
217 Thesaurus::Meanings Thesaurus::lookup(docstring const & t, docstring const & lang)
222 if (!d->addThesaurus(lang))
225 for (Thesauri::const_iterator it = d->thes_.begin();
226 it != d->thes_.end(); ++it) {
227 if (it->first == lang) {
236 string const encoding = mythes->get_th_encoding();
239 string const text = to_iconv_encoding(support::lowercase(t), encoding);
240 int len = strlen(text.c_str());
241 int count = mythes->Lookup(text.c_str(), len, &pmean);
245 // don't change value of pmean or count
246 // they are needed for the CleanUpAfterLookup routine
250 for (int i = 0; i < count; i++) {
251 meaning = from_iconv_encoding(string(pm->defn), encoding);
253 if (support::prefixIs(meaning, '-'))
254 meaning = support::ltrim(meaning, "- ");
255 for (int j = 0; j < pm->count; j++) {
256 ret = from_iconv_encoding(string(pm->psyns[j]), encoding);
258 meanings[meaning].push_back(ret);
261 // now clean up all allocated memory
262 mythes->CleanUpAfterLookup(&pmean, count);
264 for (Meanings::iterator it = meanings.begin();
265 it != meanings.end(); ++it)
266 sort(it->second.begin(), it->second.end());
273 struct Thesaurus::Private
278 Thesaurus::Meanings Thesaurus::lookup(docstring const &, docstring const &)
284 bool Thesaurus::thesaurusAvailable(docstring const &) const
289 #endif // HAVE_LIBMYTHES
290 #endif // HAVE_LIBAIKSAURUS
292 Thesaurus::Thesaurus() : d(new Thesaurus::Private)
297 Thesaurus::~Thesaurus()