Ambiguous decomposition sequences now decompose to first ranked decomposition

(resolves #18)

ZRColaCompile/dbsource.cpp

@@ -302,7 +302,7 @@ bool ZRCola::DBSource::SelectTranslations(ATL::CComPtr<ADORecordset> &rs) const
     wxCHECK(SUCCEEDED(::CoCreateInstance(CLSID_CADORecordset, NULL, CLSCTX_ALL, IID_IADORecordset, (LPVOID*)&rs)), false);
 
     // Open it.
-    if (FAILED(rs->Open(ATL::CComVariant(L"SELECT [komb], [znak] FROM [VRS_ReplChar] WHERE [rang_komb]=1"), ATL::CComVariant(m_db), adOpenStatic, adLockReadOnly, adCmdText))) {
+    if (FAILED(rs->Open(ATL::CComVariant(L"SELECT [komb], [znak], [rang_znak] FROM [VRS_ReplChar] WHERE [rang_komb]=1"), ATL::CComVariant(m_db), adOpenStatic, adLockReadOnly, adCmdText))) {
         _ftprintf(stderr, wxT("%s: error ZCC0040: Error loading compositions from database. Please make sure the file is ZRCola.zrc compatible.\n"), m_filename.c_str());
         LogErrors();
         return false;
@@ -331,6 +331,12 @@ bool ZRCola::DBSource::GetTranslation(const ATL::CComPtr<ADORecordset>& rs, ZRCo
         wxCHECK(GetUnicodeCharacter(f, t.chr), false);
     }
 
+    {
+        ATL::CComPtr<ADOField> f;
+        wxVERIFY(SUCCEEDED(flds->get_Item(ATL::CComVariant(L"rang_znak"), &f)));
+        wxCHECK(GetValue(f, t.rank), false);
+    }
+
     return true;
 }
 

ZRColaCompile/dbsource.h

@@ -41,6 +41,7 @@ namespace ZRCola {
         public:
             wchar_t chr;        ///< Composed character
             std::wstring str;   ///< Decomposed string
+            int rank;           ///< Decomposition rank
         };
 
 

ZRColaCompile/main.cpp

@@ -346,6 +346,8 @@ int _tmain(int argc, _TCHAR *argv[])
                         // Add translation to index and data.
                         unsigned __int32 idx = db.data.size();
                         db.data.push_back(trans.chr);
+                        wxASSERT_MSG((int)0xffff8000 <= trans.rank && trans.rank <= (int)0x00007fff, wxT("transformation rank out of bounds"));
+                        db.data.push_back((unsigned __int16)trans.rank);
                         std::wstring::size_type n = trans.str.length();
                         wxASSERT_MSG(n <= 0xffff, wxT("transformation string too long"));
                         db.data.push_back((unsigned __int16)n);

lib/libZRCola/include/zrcola/translate.h

@@ -46,6 +46,7 @@ namespace ZRCola {
         ///
         struct translation {
             wchar_t chr;                ///< Composed character
+            unsigned __int16 rank;      ///< Decomposition rank
             unsigned __int16 str_len;   ///< \c str length (in characters)
             wchar_t str[];              ///< Decomposed string
 
@@ -183,6 +184,9 @@ namespace ZRCola {
                      if (a.chr < b.chr) return -1;
                 else if (a.chr > b.chr) return +1;
 
+                     if (a.rank < b.rank) return -1;
+                else if (a.rank > b.rank) return +1;
+
                 int r = translation::CompareString(a.str, a.str_len, b.str, b.str_len);
                 if (r != 0) return r;
 

lib/libZRCola/src/translate.cpp

@@ -131,6 +131,16 @@ void ZRCola::translation_db::Decompose(_In_z_count_(inputMax) const wchar_t* inp
                 else if (decompSrc < c) l = m + 1;
                 else {
                     // Character found.
+
+                    // Narrow the search area on the left to start at the first decomposition in the run (first by rank).
+                    for (size_t rr = m; l < rr;) {
+                        size_t m = (l + rr) / 2;
+                        const translation &trans = idxDecomp[m];
+                        wchar_t decompSrc = trans.chr;
+                        if (c <= decompSrc) rr = m; else l = m + 1;
+                    }
+
+                    const translation &trans = idxDecomp[l];
                     if (trans.str_len && trans.str[0] != L'#' && (!lc_db || !lc_db->IsLocalCharacter(c, lang))) {
                         // Append decomposed sequence.
                         output.append(trans.str, trans.str_len);