]> git.lyx.org Git - lyx.git/blobdiff - src/Thesaurus.cpp
fix bug #6584
[lyx.git] / src / Thesaurus.cpp
index f3814baaf7adc4bf246f4ca3c52ea715589063c9..bf657762c1c2d4a423d3998a7dc8245734d3e132 100644 (file)
 #include "LyXRC.h"
 
 #include "support/FileNameList.h"
+#include "support/Package.h"
 #include "support/debug.h"
 #include "support/filetools.h"
 #include "support/gettext.h"
 #include "support/lstrings.h"
 #include "support/os.h"
-#include "support/unicode.h"
 
 #include "support/mythes/mythes.hxx"
 
@@ -29,6 +29,7 @@
 
 #include <algorithm>
 #include <cstring>
+#include <fstream>
 
 using namespace std;
 using namespace lyx::support;
@@ -38,26 +39,10 @@ namespace lyx {
 
 namespace {
 
-string const to_iconv_encoding(docstring const & s, string const & encoding)
-{
-       std::vector<char> const encoded =
-               ucs4_to_eightbit(s.data(), s.length(), encoding);
-       return string(encoded.begin(), encoded.end());
-}
-
-
-docstring const from_iconv_encoding(string const & s, string const & encoding)
-{
-       std::vector<char_type> const ucs4 =
-               eightbit_to_ucs4(s.data(), s.length(), encoding);
-       return docstring(ucs4.begin(), ucs4.end());
-}
-
 typedef std::map<docstring, MyThes *> Thesauri;
 
 } // namespace anon
 
-
 struct Thesaurus::Private
 {
        ~Private()
@@ -82,56 +67,100 @@ struct Thesaurus::Private
        ///
        typedef std::pair<std::string, std::string> ThesFiles;
        ///
+       ThesFiles getThesaurus(string const & path, docstring const & lang);
        ThesFiles getThesaurus(docstring const & lang);
        /// add a thesaurus to the list
        bool addThesaurus(docstring const & lang);
 
        /// the thesauri
        Thesauri thes_;
+
+       /// the location below system/user directory
+       /// there the data+idx files lookup will happen
+       const string dataDirectory(void) { return "thes"; }
+
 };
 
 
-pair<string, string> Thesaurus::Private::getThesaurus(docstring const & lang)
+pair<string,string> Thesaurus::Private::getThesaurus(string const & path, docstring const & lang)
 {
-       string const thes_path = external_path(lyxrc.thesaurusdir_path);
-       LYXERR(Debug::FILES, "thesaurus path: " << thes_path);
-       if (thes_path.empty())
+       FileName base(path);
+       if (!base.isDirectory()) {
                return make_pair(string(), string());
-
-       if (thesaurusAvailable(lang))
-               return make_pair(string(), string());
-
-       FileNameList const idx_files = FileName(thes_path).dirList("idx");
-       FileNameList const data_files = FileName(thes_path).dirList("dat");
+       }
+       FileNameList const idx_files = base.dirList("idx");
+       FileNameList const data_files = base.dirList("dat");
        string idx;
        string data;
-
-       for (FileNameList::const_iterator it = idx_files.begin();
-            it != idx_files.end(); ++it) {
-               LYXERR(Debug::FILES, "found thesaurus idx file: " << it->onlyFileName());
-               if (contains(it->onlyFileName(), to_ascii(lang))) {
-                       idx = it->absFilename();
+       string basename;
+
+       LYXERR(Debug::FILES, "thesaurus path: " << path);
+       for (FileNameList::const_iterator it = idx_files.begin(); it != idx_files.end(); ++it) {
+               basename = it->onlyFileNameWithoutExt();
+               if (contains(basename, to_ascii(lang))) {
+                       ifstream ifs(it->absFileName().c_str());
+                       if (ifs) {
+                               // check for appropriate version of index file
+                               string encoding; // first line is encoding
+                               int items = 0;   // second line is no. of items
+                               getline(ifs,encoding);
+                               ifs >> items;
+                               if (ifs.fail()) {
+                                       LYXERR(Debug::FILES, "ignore irregular thesaurus idx file: " << it->absFileName());
+                                       continue;
+                               }
+                               if (encoding.length() == 0 || encoding.find_first_of(',') != string::npos) {
+                                       LYXERR(Debug::FILES, "ignore version1 thesaurus idx file: " << it->absFileName());
+                                       continue;
+                               }
+                       }
+                       idx = it->absFileName();
                        LYXERR(Debug::FILES, "selected thesaurus idx file: " << idx);
                        break;
-                       }
                }
-
-       for (support::FileNameList::const_iterator it = data_files.begin();
-            it != data_files.end(); ++it) {
-               LYXERR(Debug::FILES, "found thesaurus data file: " << it->onlyFileName());
-               if (contains(it->onlyFileName(), to_ascii(lang))) {
-                       data = it->absFilename();
+       }
+       if (idx.empty()) {
+               return make_pair(string(), string());
+       }
+       for (support::FileNameList::const_iterator it = data_files.begin(); it != data_files.end(); ++it) {
+               if (contains(it->onlyFileName(), basename)) {
+                       data = it->absFileName();
                        LYXERR(Debug::FILES, "selected thesaurus data file: " << data);
                        break;
-                       }
                }
-
+       }
        return make_pair(idx, data);
 }
 
 
+pair<string,string> Thesaurus::Private::getThesaurus(docstring const & lang)
+{
+       string const thes_path = external_path(lyxrc.thesaurusdir_path);
+       pair<string,string> result ;
+
+       if (thesaurusAvailable(lang))
+               return make_pair(string(), string());
+
+       if (!thes_path.empty()) {
+               result = getThesaurus(thes_path, lang);
+       }
+       if (result.first.empty() || result.second.empty()) {
+               string const sys_path = external_path(addName(lyx::support::package().system_support().absFileName(),dataDirectory())) ;
+               result = getThesaurus(sys_path, lang);
+       }
+       if (result.first.empty() || result.second.empty()) {
+               string const user_path = external_path(addName(lyx::support::package().user_support().absFileName(),dataDirectory())) ;
+               result = getThesaurus(user_path, lang);
+       }
+       return result;
+}
+
+
 bool Thesaurus::Private::addThesaurus(docstring const & lang)
 {
+       if (thesaurusAvailable(lang))
+               return true;
+
        ThesFiles files = getThesaurus(lang);
        string const idx = files.first;
        string const data = files.second;
@@ -154,6 +183,8 @@ bool Thesaurus::thesaurusAvailable(docstring const & lang) const
 
 bool Thesaurus::thesaurusInstalled(docstring const & lang) const
 {
+       if (thesaurusAvailable(lang))
+               return true;
        pair<string, string> files = d->getThesaurus(lang);
        return (!files.first.empty() && !files.second.empty());
 }
@@ -191,17 +222,17 @@ Thesaurus::Meanings Thesaurus::lookup(docstring const & t, docstring const & lan
        // they are needed for the CleanUpAfterLookup routine
        mentry * pm = pmean;
        docstring meaning;
-       docstring ret;
        for (int i = 0; i < count; i++) {
+               vector<docstring> ret;
                meaning = from_iconv_encoding(string(pm->defn), encoding);
                // remove silly item
                if (support::prefixIs(meaning, '-'))
                        meaning = support::ltrim(meaning, "- ");
                for (int j = 0; j < pm->count; j++) {
-                       ret = from_iconv_encoding(string(pm->psyns[j]), encoding);
+                       ret.push_back(from_iconv_encoding(string(pm->psyns[j]), encoding));
                }
-       meanings[meaning].push_back(ret);
-       pm++;
+               meanings[meaning] = ret;
+               ++pm;
        }
        // now clean up all allocated memory
        mythes->CleanUpAfterLookup(&pmean, count);