added language read support for stardict format

added smart language extraction from filename to LangCoder

src/langcoder.cc

@@ -12,7 +12,7 @@ LangCoder::LangCoder()
   LangStruct ls;
   for (int i = 0; true; i++) {
     const LangCode &lc = LangCodes[i];
-    if (lc.lang.isEmpty())
+    if (lc.lang[0] == 0)
       break;
     //ls.order = i;
     //ls.icon = QIcon(":/flags/" + QString(lc.code) + ".png");
@@ -23,8 +23,8 @@ LangCoder::LangCoder()
 QString LangCoder::decode(quint32 code)
 {
   // temp!
-  if (codeMap.contains(code))
-    return LangCodes[codeMap[code]].lang;
+  if (langCoder.codeMap.contains(code))
+    return LangCodes[langCoder.codeMap[code]].lang;
 
   return QString();
 }
@@ -72,6 +72,50 @@ quint32 LangCoder::findIdForLanguage( gd::wstring const & lang )
   return 0;
 }
 
+quint32 LangCoder::guessId( const QString & lang )
+{
+  QString lstr = lang.simplified().toLower();
+
+  // too small to guess
+  if (lstr.size() < 2)
+    return 0;
+
+  // check if it could be the whole language name
+  if (lstr.size() >= 3)
+  {
+    for( LangCode const * lc = LangCodes; lc->code[ 0 ]; ++lc )
+    {
+      if ( lstr == QString( lc->lang ) )
+      {
+        // We've got a match
+        return code2toInt( lc->code );
+      }
+    }
+  }
+
+  // still not found - try to match by 2-symbol code
+  return code2toInt( lstr.left(2).toAscii().data() );
+}
+
+QPair<quint32,quint32> LangCoder::findIdsForFilename( QString const & name )
+{
+  QString nameFolded = QFileInfo( name ).fileName().toCaseFolded();
+
+  QRegExp reg( "[-_.]([a-z]{2,3})-([a-z]{2,3})[-_.]" ); reg.setMinimal(true);
+  int off = 0;
+  while ( reg.indexIn( nameFolded, off ) >= 0 )
+  {
+    quint32 from = guessId( reg.cap(1) );
+    quint32 to = guessId( reg.cap(2) );
+    if (from && to)
+      return QPair<quint32,quint32>(from, to);
+
+    off += reg.matchedLength();
+  }
+
+  return QPair<quint32,quint32>(0, 0);
+}
+
 /*
 LangStruct& LangCoder::CodeToLangStruct(const QString &code)
 {

src/langcoder.hh

@@ -7,7 +7,7 @@
 struct LangCode
 {
     char code[ 3 ]; // ISO 639-1
-    QString lang; // Language name in English
+    char *lang; // Language name in English
 };
 
 // Language codes
@@ -230,9 +230,15 @@ public:
   /// is case- and punctuation insensitive.
   static quint32 findIdForLanguage( gd::wstring const & );
 
-  //const QMap<quint32, int>& codes() { return codeMap; }
 
-  QString decode(quint32 code);
+  static QPair<quint32,quint32> findIdsForFilename( QString const & );
+
+  static quint32 guessId( const QString & lang );
+
+  /// Returns decoded name of language or empty string if not found.
+  static QString decode(quint32 code);
+
+  //const QMap<quint32, int>& codes() { return codeMap; }
 
   LangStruct langStruct(quint32 code);
 

src/stardict.cc

@@ -9,6 +9,8 @@
 #include "dictzip.h"
 #include "xdxf2html.hh"
 #include "htmlescape.hh"
+#include "langcoder.hh"
+
 #include <zlib.h>
 #include <map>
 #include <set>
@@ -69,7 +71,7 @@ struct Ifo
 enum
 {
   Signature = 0x58444953, // SIDX on little-endian, XDIS on big-endian
-  CurrentFormatVersion = 5 + BtreeIndexing::FormatVersion + Folding::Version
+  CurrentFormatVersion = 6 + BtreeIndexing::FormatVersion + Folding::Version
 };
 
 struct IdxHeader
@@ -83,6 +85,8 @@ struct IdxHeader
   uint32_t synWordCount; // Saved from Ifo::synwordcount
   uint32_t bookNameSize; // Book name's length. Used to read it then.
   uint32_t sameTypeSequenceSize; // That string's size. Used to read it then.
+  uint32_t langFrom;  // Source language
+  uint32_t langTo;    // Target language
 } __attribute__((packed));
 
 bool indexIsOldOrBad( string const & indexFile )
@@ -130,6 +134,12 @@ public:
   virtual QIcon getIcon() throw()
   { return QIcon(":/icons/icon32_stardict.png"); }
 
+  inline virtual quint32 getLangFrom() const
+  { return idxHeader.langFrom; }
+
+  inline virtual quint32 getLangTo() const
+  { return idxHeader.langTo; }
+
   virtual sptr< Dictionary::WordSearchRequest > findHeadwordsForSynonym( wstring const & )
     throw( std::exception );
 
@@ -1068,6 +1078,11 @@ vector< sptr< Dictionary::Class > > makeDictionaries(
         idxHeader.bookNameSize = ifo.bookname.size();
         idxHeader.sameTypeSequenceSize = ifo.sametypesequence.size();
 
+        QPair<quint32,quint32> langs =
+            LangCoder::findIdsForFilename( QString::fromStdString( dictFileName ) );
+        idxHeader.langFrom = langs.first;
+        idxHeader.langTo = langs.second;
+
         idx.rewind();
 
         idx.write( &idxHeader, sizeof( idxHeader ) );