Bug 705910 - Indexing and searching cannot treat non ASCII identifiers

qtools/Doxyfile

@@ -174,7 +174,7 @@ QHP_SECT_FILTER_ATTRS  =
 QHG_LOCATION           =
 GENERATE_ECLIPSEHELP   = YES
 ECLIPSE_DOC_ID         = org.doxygen.qtools
-DISABLE_INDEX          = YES
+DISABLE_INDEX          = NO
 GENERATE_TREEVIEW      = YES
 ENUM_VALUES_PER_LINE   = 4
 TREEVIEW_WIDTH         = 250

src/index.cpp

@@ -119,23 +119,28 @@ static void endIndexHierarchy(OutputList &ol,int level)
 class MemberIndexList : public QList<MemberDef>
 {
   public:
-    MemberIndexList() : QList<MemberDef>() {}
+    typedef MemberDef ElementType;
+    MemberIndexList(uint letter) : QList<MemberDef>(), m_letter(letter) {}
     ~MemberIndexList() {}
     int compareItems(QCollection::Item item1, QCollection::Item item2)
     {
       MemberDef *md1=(MemberDef *)item1;
       MemberDef *md2=(MemberDef *)item2;
-      return qstricmp(md1->name(),md2->name());
+      int result = qstricmp(md1->name(),md2->name());
+      if (result==0)
+      {
+        result = qstricmp(md1->qualifiedName(),md2->qualifiedName());
+      }
+      return result;
     }
+    uint letter() const { return m_letter; }
+  private:
+    uint m_letter;
 };
 
-#define MEMBER_INDEX_ENTRIES 256
-
-static MemberIndexList g_memberIndexLetterUsed[CMHL_Total][MEMBER_INDEX_ENTRIES];
-static MemberIndexList g_fileIndexLetterUsed[FMHL_Total][MEMBER_INDEX_ENTRIES];
-static MemberIndexList g_namespaceIndexLetterUsed[NMHL_Total][MEMBER_INDEX_ENTRIES];
-
-//static bool g_classIndexLetterUsed[CHL_Total][256];
+static LetterToIndexMap<MemberIndexList> g_memberIndexLetterUsed[CMHL_Total];
+static LetterToIndexMap<MemberIndexList> g_fileIndexLetterUsed[FMHL_Total];
+static LetterToIndexMap<MemberIndexList> g_namespaceIndexLetterUsed[NMHL_Total];
 
 const int maxItemsBeforeQuickIndex = MAX_ITEMS_BEFORE_QUICK_INDEX;
 
@@ -1690,21 +1695,39 @@ static void writeAnnotatedClassList(OutputList &ol)
   ol.endIndexList();
 }
 
-static QCString letterToLabel(char startLetter)
+static QCString letterToLabel(uint startLetter)
 {
-  QCString s(5); 
-  if (isId(startLetter))
+  char s[10];
+  if (startLetter>0x20 && startLetter<=0x7f) // printable ASCII character
   {
-    s[0]=startLetter; s[1]=0;
+    s[0]=(char)startLetter;
+    s[1]=0;
   }
   else
   {
     const char hex[]="0123456789abcdef";
-    s[0]='0';
-    s[1]='x';
-    s[2]=hex[startLetter>>4];
-    s[3]=hex[startLetter&0xF];
-    s[4]=0;
+    int i=0;
+    s[i++]='0';
+    s[i++]='x';
+    if (startLetter>(1<<24)) // 4 byte character
+    {
+      s[i++]=hex[(startLetter>>28)&0xf];
+      s[i++]=hex[(startLetter>>24)&0xf];
+    }
+    if (startLetter>(1<<16)) // 3 byte character
+    {
+      s[i++]=hex[(startLetter>>20)&0xf];
+      s[i++]=hex[(startLetter>>16)&0xf];
+    }
+    if (startLetter>(1<<8)) // 2 byte character
+    {
+      s[i++]=hex[(startLetter>>12)&0xf];
+      s[i++]=hex[(startLetter>>8)&0xf];
+    }
+    // one byte character
+    s[i++]=hex[(startLetter>>4)&0xf];
+    s[i++]=hex[(startLetter>>0)&0xf];
+    s[i++]=0;
   }
   return s;
 }
@@ -1714,7 +1737,9 @@ static QCString letterToLabel(char startLetter)
 /** Special class list where sorting takes IGNORE_PREFIX into account. */
 class PrefixIgnoreClassList : public ClassList
 {
-public:
+  public:
+    typedef ClassDef ElementType;
+    PrefixIgnoreClassList(uint letter) : m_letter(letter) {}
     virtual int compareItems(QCollection::Item item1, QCollection::Item item2)
     {
       ClassDef *c1=(ClassDef *)item1;
@@ -1724,25 +1749,28 @@ public:
       QCString n2 = c2->className();
       return qstricmp (n1.data()+getPrefixIndex(n1), n2.data()+getPrefixIndex(n2));
     }
+    uint letter() const { return m_letter; }
+  private:
+    uint m_letter;
 };
 
 /** Class representing a cell in the alphabetical class index. */
 class AlphaIndexTableCell
 {
   public:
-    AlphaIndexTableCell(int row,int col,uchar letter,ClassDef *cd) : 
+    AlphaIndexTableCell(int row,int col,uint letter,ClassDef *cd) : 
       m_letter(letter), m_class(cd), m_row(row), m_col(col) 
     { //printf("AlphaIndexTableCell(%d,%d,%c,%s)\n",row,col,letter!=0 ? letter: '-',
       //       cd!=(ClassDef*)0x8 ? cd->name().data() : "<null>"); 
     }
 
     ClassDef *classDef() const { return m_class; }
-    uchar letter()       const { return m_letter; }
+    uint letter()        const { return m_letter; }
     int row()            const { return m_row; }
     int column()         const { return m_col; }
 
   private:
-    uchar m_letter;
+    uint m_letter;
     ClassDef *m_class;
     int m_row;
     int m_col;
@@ -1770,12 +1798,31 @@ class AlphaIndexTableColumns : public QList<AlphaIndexTableRows>
     AlphaIndexTableColumns() { setAutoDelete(TRUE); }
 };
 
+class UsedIndexLetters : public SIntDict<uint>
+{
+  public:
+    UsedIndexLetters() : SIntDict<uint>(257) { setAutoDelete(TRUE); }
+    int compareItems( QCollection::Item item1, QCollection::Item item2)
+    {
+      int *p1=(int *)item1;
+      int *p2=(int *)item2;
+      return *p1 - *p2; // subtracting is done by int not uint.
+    }
+    void add(uint letter)
+    {
+      uint *v = find(letter);
+      if (v==0)
+      {
+        append(letter,new uint(letter));
+      }
+    }
+};
+
 // write an alphabetical index of all class with a header for each letter
 static void writeAlphabeticalClassList(OutputList &ol)
 {
   // What starting letters are used
-  bool indexLetterUsed[256];
-  memset (indexLetterUsed, 0, sizeof (indexLetterUsed));
+  UsedIndexLetters indexLettersUsed;
 
   // first count the number of headers
   ClassSDict::Iterator cli(*Doxygen::classSDict);
@@ -1791,24 +1838,25 @@ static void writeAlphabeticalClassList(OutputList &ol)
 	     
       int index = getPrefixIndex(cd->className());
       //printf("name=%s index=%d %d\n",cd->className().data(),index,cd->protection());
-      startLetter=toupper(cd->className().at(index))&0xFF;
-      indexLetterUsed[startLetter] = true;
+      startLetter=getUtf8CodeToUpper(cd->className(),index);
+      indexLettersUsed.add(startLetter);
     }
   }
+  indexLettersUsed.sort();
 
   // write quick link index (row of letters)
   QCString alphaLinks = "<div class=\"qindex\">";
-  int l;
-  for (l=0; l<256; l++)
-  {
-    if (indexLetterUsed[l])
+  SIntDict<uint>::Iterator it(indexLettersUsed);
+  uint *pLetter;
+  for (it.toFirst();(pLetter=it.current());++it)
   {
     if (headerItems) alphaLinks += "&#160;|&#160;";
     headerItems++;
+    QCString li = letterToLabel(*pLetter);
+    QCString ls = QString(QChar(*pLetter)).utf8();
     alphaLinks += (QCString)"<a class=\"qindex\" href=\"#letter_" + 
-                    (char)l + "\">" + 
-                    (char)l + "</a>";
-    }
+                  li + "\">" + 
+                  ls + "</a>";
   }
   alphaLinks += "</div>\n";
   ol.writeString(alphaLinks);
@@ -1825,7 +1873,7 @@ static void writeAlphabeticalClassList(OutputList &ol)
   //    headerItems,totalItems,columns,rows,itemsInLastRow);
 
   // Keep a list of classes for each starting letter
-  PrefixIgnoreClassList classesByLetter[256];
+  LetterToIndexMap<PrefixIgnoreClassList> classesByLetter;
   AlphaIndexTableColumns tableColumns;
 
   // fill the columns with the class list (row elements in each column,
@@ -1841,16 +1889,20 @@ static void writeAlphabeticalClassList(OutputList &ol)
     if (cd->isLinkableInProject() && cd->templateMaster()==0)
     {
       int index = getPrefixIndex(cd->className());
-      startLetter=toupper(cd->className().at(index))&0xFF;
+      startLetter=getUtf8Code(cd->className(),index);
       // Do some sorting again, since the classes are sorted by name with 
       // prefix, which should be ignored really.
       if (cd->getLanguage()==SrcLangExt_VHDL)
       {
         if ((VhdlDocGen::VhdlClasses)cd->protection()==VhdlDocGen::ENTITYCLASS )// no architecture
-          classesByLetter[startLetter].inSort(cd);
+        {
+          classesByLetter.append(startLetter,cd);
+        }
       }
       else
-      classesByLetter[startLetter].inSort(cd);
+      {
+        classesByLetter.append(startLetter,cd);
+      }
     }
   }
 
@@ -1871,27 +1923,31 @@ static void writeAlphabeticalClassList(OutputList &ol)
   AlphaIndexTableRows *tableRows = new AlphaIndexTableRows;
   tableColumns.append(tableRows);
   int col=0,row=0,maxRows=0;
-  for (l=0; l<256; l++)
-  {
-    if (classesByLetter[l].count()>0)
+  PrefixIgnoreClassList *cl;
+  SIntDict<PrefixIgnoreClassList>::Iterator lit(classesByLetter);
+  for (lit.toFirst();(cl=lit.current());++lit)
   {
+    uint l = cl->letter();
     // add special header cell
-      tableRows->append(new AlphaIndexTableCell(row,col,(uchar)l,(ClassDef*)0x8));
+    tableRows->append(new AlphaIndexTableCell(row,col,l,(ClassDef*)0x8));
     row++;
     tableRows->append(new AlphaIndexTableCell(row,col,0,(ClassDef*)0x8));
     row++;
-      tableRows->append(new AlphaIndexTableCell(row,col,0,classesByLetter[l].at(0)));
+    ClassListIterator cit(*cl);
+    cit.toFirst();
+    ClassDef *cd = cit.current();
+    ++cit;
+    tableRows->append(new AlphaIndexTableCell(row,col,0,cd));
     row++; 
     NEXT_ROW();
-      for (i=1; i<(int)classesByLetter[l].count(); i++)
+    for (;(cd=cit.current()); ++cit)
     {
       // add normal cell
-        tableRows->append(new AlphaIndexTableCell(row,col,0,classesByLetter[l].at(i)));
+      tableRows->append(new AlphaIndexTableCell(row,col,0,cd));
       row++;
       NEXT_ROW();
     }
   }
-  }
 
   // create row iterators for each column
   AlphaIndexTableRowsIterator **colIterators = new AlphaIndexTableRowsIterator*[columns];
@@ -1936,7 +1992,7 @@ static void writeAlphabeticalClassList(OutputList &ol)
               ol.writeString("<table border=\"0\" cellspacing=\"0\" cellpadding=\"0\">"
                   "<tr>"
                   "<td><div class=\"ah\">&#160;&#160;"); 
-              ol.writeString(s);
+              ol.writeString(QString(QChar(cell->letter())).utf8());
               ol.writeString(         "&#160;&#160;</div>"
                   "</td>"
                   "</tr>"
@@ -2164,14 +2220,9 @@ static void writeNamespaceLinkForMember(OutputList &ol,MemberDef *md,const char
 }
 
 static void writeMemberList(OutputList &ol,bool useSections,int page,
-                            MemberIndexList memberLists[MEMBER_INDEX_ENTRIES],
+                            const LetterToIndexMap<MemberIndexList> &memberLists,
                             DefinitionIntf::DefType type)
 {
-  int pi;
-  // page==-1 => write all member indices to one page (used when total members is small)
-  // page!=-1 => write all member for this page only (used when total member is large)
-  int startIndex = page==-1 ? 0                      : page;
-  int endIndex   = page==-1 ? MEMBER_INDEX_ENTRIES-1 : page;
   ASSERT((int)type<3);
 
   typedef void (*writeLinkForMember_t)(OutputList &ol,MemberDef *md,const char *separator,
@@ -2189,10 +2240,16 @@ static void writeMemberList(OutputList &ol,bool useSections,int page,
   bool first=TRUE;
   bool firstSection=TRUE;
   bool firstItem=TRUE;
-  for (pi=startIndex; pi<=endIndex; pi++) // page==-1 => pi=[0..127], page!=-1 => pi=page 
+  MemberIndexList *ml;
+  SIntDict<MemberIndexList>::Iterator it(memberLists);
+  for (it.toFirst();(ml=it.current());++it)
   {
-    MemberIndexList *ml = &memberLists[pi];
-    if (ml->count()==0) continue;
+    if (page!=-1)
+    {
+      ml = memberLists[page];
+      it.toLast();
+    }
+    if (ml==0 || ml->count()==0) continue;
     ml->sort();
     QListIterator<MemberDef> mli(*ml);
     MemberDef *md;
@@ -2211,10 +2268,8 @@ static void writeMemberList(OutputList &ol,bool useSections,int page,
         {
           if (!firstItem)    ol.endItemListItem();
           if (!firstSection) ol.endItemList();
-          char cl[2];
-          cl[0] = tolower(name.at(startIndex));
-          cl[1] = 0;
-          QCString cs = letterToLabel(cl[0]);
+          QCString cs = letterToLabel(ml->letter());
+          QCString cl = QString(QChar(ml->letter())).utf8();
           QCString anchor=(QCString)"index_"+cs;
           QCString title=(QCString)"- "+cl+" -";
           ol.startSection(anchor,title,SectionInfo::Subsection);
@@ -2260,15 +2315,11 @@ static void writeMemberList(OutputList &ol,bool useSections,int page,
 
 void initClassMemberIndices()
 {
-  int i=0;
   int j=0;
   for (j=0;j<CMHL_Total;j++)
   {
     documentedClassMembers[j]=0;
-    for (i=0;i<MEMBER_INDEX_ENTRIES;i++) 
-    {
-      g_memberIndexLetterUsed[j][i].clear();
-    }
+    g_memberIndexLetterUsed[j].clear();
   }
 }
 
@@ -2286,8 +2337,7 @@ void addClassMemberNameToIndex(MemberDef *md)
   {
     QCString n = md->name();
     int index = getPrefixIndex(n);
-    uchar charCode = (uchar)n.at(index);
-    uint letter = charCode<128 ? tolower(charCode) : charCode;
+    uint letter = getUtf8CodeToLower(n,index);
     if (!n.isEmpty()) 
     {
       bool isFriendToHide = hideFriendCompounds &&
@@ -2296,48 +2346,48 @@ void addClassMemberNameToIndex(MemberDef *md)
          QCString(md->typeString())=="friend union");
       if (!(md->isFriend() && isFriendToHide))
       {
-        g_memberIndexLetterUsed[CMHL_All][letter].append(md);
+        g_memberIndexLetterUsed[CMHL_All].append(letter,md);
         documentedClassMembers[CMHL_All]++;
       }
       if (md->isFunction()  || md->isSlot() || md->isSignal())
       {
-        g_memberIndexLetterUsed[CMHL_Functions][letter].append(md);
+        g_memberIndexLetterUsed[CMHL_Functions].append(letter,md);
         documentedClassMembers[CMHL_Functions]++;
       } 
       else if (md->isVariable())
       {
-        g_memberIndexLetterUsed[CMHL_Variables][letter].append(md);
+        g_memberIndexLetterUsed[CMHL_Variables].append(letter,md);
         documentedClassMembers[CMHL_Variables]++;
       }
       else if (md->isTypedef())
       {
-        g_memberIndexLetterUsed[CMHL_Typedefs][letter].append(md);
+        g_memberIndexLetterUsed[CMHL_Typedefs].append(letter,md);
         documentedClassMembers[CMHL_Typedefs]++;
       }
       else if (md->isEnumerate())
       {
-        g_memberIndexLetterUsed[CMHL_Enums][letter].append(md);
+        g_memberIndexLetterUsed[CMHL_Enums].append(letter,md);
         documentedClassMembers[CMHL_Enums]++;
       }
       else if (md->isEnumValue())
       {
-        g_memberIndexLetterUsed[CMHL_EnumValues][letter].append(md);
+        g_memberIndexLetterUsed[CMHL_EnumValues].append(letter,md);
         documentedClassMembers[CMHL_EnumValues]++;
       }
       else if (md->isProperty())
       {
-        g_memberIndexLetterUsed[CMHL_Properties][letter].append(md);
+        g_memberIndexLetterUsed[CMHL_Properties].append(letter,md);
         documentedClassMembers[CMHL_Properties]++;
       }
       else if (md->isEvent())
       {
-        g_memberIndexLetterUsed[CMHL_Events][letter].append(md);
+        g_memberIndexLetterUsed[CMHL_Events].append(letter,md);
         documentedClassMembers[CMHL_Events]++;
       }
       else if (md->isRelated() || md->isForeign() ||
                (md->isFriend() && !isFriendToHide))
       {
-        g_memberIndexLetterUsed[CMHL_Related][letter].append(md);
+        g_memberIndexLetterUsed[CMHL_Related].append(letter,md);
         documentedClassMembers[CMHL_Related]++;
       }
     }
@@ -2348,15 +2398,11 @@ void addClassMemberNameToIndex(MemberDef *md)
 
 void initNamespaceMemberIndices()
 {
-  int i=0;
   int j=0;
   for (j=0;j<NMHL_Total;j++)
   {
     documentedNamespaceMembers[j]=0;
-    for (i=0;i<MEMBER_INDEX_ENTRIES;i++) 
-    {
-      g_namespaceIndexLetterUsed[j][i].clear();
-    }
+    g_namespaceIndexLetterUsed[j].clear();
   }
 }
 
@@ -2367,36 +2413,35 @@ void addNamespaceMemberNameToIndex(MemberDef *md)
   {
     QCString n = md->name();
     int index = getPrefixIndex(n);
-    uchar charCode = (uchar)n.at(index);
-    uint letter = charCode<128 ? tolower(charCode) : charCode;
+    uint letter = getUtf8CodeToLower(n,index);
     if (!n.isEmpty()) 
     {
-      g_namespaceIndexLetterUsed[NMHL_All][letter].append(md);
+      g_namespaceIndexLetterUsed[NMHL_All].append(letter,md);
       documentedNamespaceMembers[NMHL_All]++;
 
       if (md->isFunction()) 
       {
-        g_namespaceIndexLetterUsed[NMHL_Functions][letter].append(md);
+        g_namespaceIndexLetterUsed[NMHL_Functions].append(letter,md);
         documentedNamespaceMembers[NMHL_Functions]++;
       }
       else if (md->isVariable()) 
       {
-        g_namespaceIndexLetterUsed[NMHL_Variables][letter].append(md);
+        g_namespaceIndexLetterUsed[NMHL_Variables].append(letter,md);
         documentedNamespaceMembers[NMHL_Variables]++;
       }
       else if (md->isTypedef())
       {
-        g_namespaceIndexLetterUsed[NMHL_Typedefs][letter].append(md);
+        g_namespaceIndexLetterUsed[NMHL_Typedefs].append(letter,md);
         documentedNamespaceMembers[NMHL_Typedefs]++;
       }
       else if (md->isEnumerate())
       {
-        g_namespaceIndexLetterUsed[NMHL_Enums][letter].append(md);
+        g_namespaceIndexLetterUsed[NMHL_Enums].append(letter,md);
         documentedNamespaceMembers[NMHL_Enums]++;
       }
       else if (md->isEnumValue())
       {
-        g_namespaceIndexLetterUsed[NMHL_EnumValues][letter].append(md);
+        g_namespaceIndexLetterUsed[NMHL_EnumValues].append(letter,md);
         documentedNamespaceMembers[NMHL_EnumValues]++;
       }
     }
@@ -2407,15 +2452,11 @@ void addNamespaceMemberNameToIndex(MemberDef *md)
 
 void initFileMemberIndices()
 {
-  int i=0;
   int j=0;
   for (j=0;j<NMHL_Total;j++)
   {
     documentedFileMembers[j]=0;
-    for (i=0;i<MEMBER_INDEX_ENTRIES;i++) 
-    {
-      g_fileIndexLetterUsed[j][i].clear();
-    }
+    g_fileIndexLetterUsed[j].clear();
   }
 }
 
@@ -2426,41 +2467,40 @@ void addFileMemberNameToIndex(MemberDef *md)
   {
     QCString n = md->name();
     int index = getPrefixIndex(n);
-    uchar charCode = (uchar)n.at(index);
-    uint letter = charCode<128 ? tolower(charCode) : charCode;
+    uint letter = getUtf8CodeToLower(n,index);
     if (!n.isEmpty()) 
     {
-      g_fileIndexLetterUsed[FMHL_All][letter].append(md);
+      g_fileIndexLetterUsed[FMHL_All].append(letter,md);
       documentedFileMembers[FMHL_All]++;
 
       if (md->isFunction()) 
       {
-        g_fileIndexLetterUsed[FMHL_Functions][letter].append(md);
+        g_fileIndexLetterUsed[FMHL_Functions].append(letter,md);
         documentedFileMembers[FMHL_Functions]++;
       }
       else if (md->isVariable()) 
       {
-        g_fileIndexLetterUsed[FMHL_Variables][letter].append(md);
+        g_fileIndexLetterUsed[FMHL_Variables].append(letter,md);
         documentedFileMembers[FMHL_Variables]++;
       }
       else if (md->isTypedef())
       {
-        g_fileIndexLetterUsed[FMHL_Typedefs][letter].append(md);
+        g_fileIndexLetterUsed[FMHL_Typedefs].append(letter,md);
         documentedFileMembers[FMHL_Typedefs]++;
       }
       else if (md->isEnumerate())
       {
-        g_fileIndexLetterUsed[FMHL_Enums][letter].append(md);
+        g_fileIndexLetterUsed[FMHL_Enums].append(letter,md);
         documentedFileMembers[FMHL_Enums]++;
       }
       else if (md->isEnumValue())
       {
-        g_fileIndexLetterUsed[FMHL_EnumValues][letter].append(md);
+        g_fileIndexLetterUsed[FMHL_EnumValues].append(letter,md);
         documentedFileMembers[FMHL_EnumValues]++;
       }
       else if (md->isDefine())
       {
-        g_fileIndexLetterUsed[FMHL_Defines][letter].append(md);
+        g_fileIndexLetterUsed[FMHL_Defines].append(letter,md);
         documentedFileMembers[FMHL_Defines]++;
       }
     }
@@ -2470,18 +2510,18 @@ void addFileMemberNameToIndex(MemberDef *md)
 //----------------------------------------------------------------------------
 
 static void writeQuickMemberIndex(OutputList &ol,
-    MemberIndexList charUsed[MEMBER_INDEX_ENTRIES],int page,
+    const LetterToIndexMap<MemberIndexList> &charUsed,uint page,
     QCString fullName,bool multiPage)
 {
   bool first=TRUE;
-  int i;
   startQuickIndexList(ol,TRUE);
-  for (i=33;i<127;i++)
-  {
-    char is[2];is[0]=(char)i;is[1]='\0';
-    QCString ci = letterToLabel((char)i);
-    if (charUsed[i].count()>0)
+  SIntDict<MemberIndexList>::Iterator it(charUsed);
+  MemberIndexList *ml;
+  for (it.toFirst();(ml=it.current());++it)
   {
+    uint i = ml->letter();
+    QCString is = letterToLabel(i);
+    QCString ci = QString(QChar(i)).utf8();
     QCString anchor;
     QCString extension=Doxygen::htmlFileExtension;
     if (!multiPage)
@@ -2489,13 +2529,12 @@ static void writeQuickMemberIndex(OutputList &ol,
     else if (first) 
       anchor=fullName+extension+"#index_";
     else 
-        anchor=fullName+QCString().sprintf("_0x%02x",i)+extension+"#index_";
+      anchor=fullName+"_"+letterToLabel(i)+extension+"#index_";
     startQuickIndexItem(ol,anchor+ci,i==page,TRUE,first);
     ol.writeString(is);
     endQuickIndexItem(ol);
     first=FALSE;
   }
-  }
   endQuickIndexList(ol);
 }
 
@@ -2538,11 +2577,9 @@ static void writeClassMemberIndexFiltered(OutputList &ol, ClassMemberHighlight h
   static bool disableIndex     = Config_getBool("DISABLE_INDEX");
 
   bool multiPageIndex=FALSE;
-  int numPages=1;
   if (documentedClassMembers[hl]>MAX_ITEMS_BEFORE_MULTIPAGE_INDEX)
   {
     multiPageIndex=TRUE;
-    numPages=127;
   }
 
   ol.pushGeneratorState();
@@ -2561,22 +2598,20 @@ static void writeClassMemberIndexFiltered(OutputList &ol, ClassMemberHighlight h
     if (multiPageIndex) Doxygen::indexList->incContentsDepth();
   }
 
-  int page;
   bool first=TRUE;
-  for (page=0;page<numPages;page++)
-  {
-    if (!multiPageIndex || g_memberIndexLetterUsed[hl][page].count()>0)
+  SIntDict<MemberIndexList>::Iterator it(g_memberIndexLetterUsed[hl]);
+  MemberIndexList *ml;
+  for (it.toFirst();(ml=it.current());++it)
   {
+    uint page = ml->letter();
     QCString fileName = getCmhlInfo(hl)->fname;
     if (multiPageIndex)
     { 
       if (!first)
       {
-          fileName+=QCString().sprintf("_0x%02x",page);
+        fileName+="_"+letterToLabel(page);
       }
-        char cs[2];
-        cs[0]=page;
-        cs[1]=0;
+      QCString cs = QString(QChar(page)).utf8();
       if (addToIndex)
       {
         Doxygen::indexList->addContentsItem(FALSE,cs,0,fileName,0,FALSE,TRUE);
@@ -2638,7 +2673,7 @@ static void writeClassMemberIndexFiltered(OutputList &ol, ClassMemberHighlight h
       // normal lists otherwise
       ol.writeString("&#160;");
     }
-      //ol.newParagraph();  // FIXME:PARA
+
     writeMemberList(ol,quickIndex,
         multiPageIndex?page:-1,
         g_memberIndexLetterUsed[hl],
@@ -2646,7 +2681,6 @@ static void writeClassMemberIndexFiltered(OutputList &ol, ClassMemberHighlight h
     endFile(ol);
     first=FALSE;
   }
-  }
 
   if (multiPageIndex && addToIndex) Doxygen::indexList->decContentsDepth();
 
@@ -2716,11 +2750,9 @@ static void writeFileMemberIndexFiltered(OutputList &ol, FileMemberHighlight hl)
   static bool disableIndex     = Config_getBool("DISABLE_INDEX");
 
   bool multiPageIndex=FALSE;
-  int numPages=1;
   if (documentedFileMembers[hl]>MAX_ITEMS_BEFORE_MULTIPAGE_INDEX)
   {
     multiPageIndex=TRUE;
-    numPages=127;
   }
 
   ol.pushGeneratorState();
@@ -2738,22 +2770,20 @@ static void writeFileMemberIndexFiltered(OutputList &ol, FileMemberHighlight hl)
     if (multiPageIndex) Doxygen::indexList->incContentsDepth();
   }
 
-  int page;
   bool first=TRUE;
-  for (page=0;page<numPages;page++)
-  {
-    if (!multiPageIndex || g_fileIndexLetterUsed[hl][page].count()>0)
+  SIntDict<MemberIndexList>::Iterator it(g_fileIndexLetterUsed[hl]);
+  MemberIndexList *ml;
+  for (it.toFirst();(ml=it.current());++it)
   {
+    uint page = ml->letter();
     QCString fileName = getFmhlInfo(hl)->fname;
     if (multiPageIndex)
     {
       if (!first)
       {
-          fileName+=QCString().sprintf("_0x%02x",page);
+        fileName+="_"+letterToLabel(page);
       }
-        char cs[2];
-        cs[0]=page;
-        cs[1]=0;
+      QCString cs = QString(QChar(page)).utf8();
       if (addToIndex)
       {
         Doxygen::indexList->addContentsItem(FALSE,cs,0,fileName,0,FALSE,TRUE);
@@ -2813,8 +2843,7 @@ static void writeFileMemberIndexFiltered(OutputList &ol, FileMemberHighlight hl)
       // normal lists otherwise
       ol.writeString("&#160;");
     }
-      //ol.newParagraph();  // FIXME:PARA
-      //writeFileMemberList(ol,quickIndex,hl,page);
+
     writeMemberList(ol,quickIndex,
         multiPageIndex?page:-1,
         g_fileIndexLetterUsed[hl],
@@ -2822,7 +2851,6 @@ static void writeFileMemberIndexFiltered(OutputList &ol, FileMemberHighlight hl)
     endFile(ol);
     first=FALSE;
   }
-  }
   if (multiPageIndex && addToIndex) Doxygen::indexList->decContentsDepth();
   ol.popGeneratorState();
 }
@@ -2890,11 +2918,9 @@ static void writeNamespaceMemberIndexFiltered(OutputList &ol,
 
 
   bool multiPageIndex=FALSE;
-  int numPages=1;
   if (documentedNamespaceMembers[hl]>MAX_ITEMS_BEFORE_MULTIPAGE_INDEX)
   {
     multiPageIndex=TRUE;
-    numPages=127;
   }
 
   ol.pushGeneratorState();
@@ -2912,22 +2938,20 @@ static void writeNamespaceMemberIndexFiltered(OutputList &ol,
     if (multiPageIndex) Doxygen::indexList->incContentsDepth();
   }
 
-  int page;
   bool first=TRUE;
-  for (page=0;page<numPages;page++)
-  {
-    if (!multiPageIndex || g_namespaceIndexLetterUsed[hl][page].count()>0)
+  SIntDict<MemberIndexList>::Iterator it(g_namespaceIndexLetterUsed[hl]);
+  MemberIndexList *ml;
+  for (it.toFirst();(ml=it.current());++it)
   {
+    uint page = ml->letter();
     QCString fileName = getNmhlInfo(hl)->fname;
     if (multiPageIndex)
     {
       if (!first)
       {
-          fileName+=QCString().sprintf("_0x%02x",page);
+        fileName+="_"+letterToLabel(page);
       }
-        char cs[2];
-        cs[0]=page;
-        cs[1]=0;
+      QCString cs = QString(QChar(page)).utf8();
       if (addToIndex)
       {
         Doxygen::indexList->addContentsItem(FALSE,cs,0,fileName,0,FALSE,TRUE);
@@ -2988,16 +3012,13 @@ static void writeNamespaceMemberIndexFiltered(OutputList &ol,
       // normal lists otherwise
       ol.writeString("&#160;");
     }
-      //ol.newParagraph(); // FIXME:PARA
 
-      //writeNamespaceMemberList(ol,quickIndex,hl,page);
     writeMemberList(ol,quickIndex,
         multiPageIndex?page:-1,
         g_namespaceIndexLetterUsed[hl],
         Definition::TypeNamespace);
     endFile(ol);
   }
-  }
   if (multiPageIndex && addToIndex) Doxygen::indexList->decContentsDepth();
   ol.popGeneratorState();
 }
@@ -3093,167 +3114,6 @@ static void writeExampleIndex(OutputList &ol)
 }
 
 
-//----------------------------------------------------------------------------
-
-template<typename T>
-bool writeMemberNavIndex(FTextStream &t,
-                         int indent,
-                         int n,
-                         int documentedMembers[],
-                         MemberIndexList indexLetterUsed[][MEMBER_INDEX_ENTRIES],
-                         const T *(*getInfo)(int),
-                         bool &first
-                        )
-
-{
-  bool found=FALSE;
-  QCString indentStr;
-  indentStr.fill(' ',indent*2);
-  // index items per category member lists
-  int i;
-  for (i=0;i<n;i++)
-  {
-    bool hasIndex       = documentedMembers[i]>0;
-    bool quickIndex     = documentedMembers[i]>maxItemsBeforeQuickIndex;
-    bool multiIndexPage = documentedMembers[i]>MAX_ITEMS_BEFORE_MULTIPAGE_INDEX;
-    if (hasIndex)
-    {
-      // terminate previous entry
-      if (!first) t << "," << endl;
-      first = FALSE;
-
-      // start entry
-      if (!found)
-      {
-        t << "[" << endl;
-      }
-      found = TRUE;
-
-      t << indentStr << "  [ ";
-      t << "\"" << fixSpaces(getInfo(i)->title) << "\", ";
-      t << "\"" << getInfo(i)->fname << Doxygen::htmlFileExtension << "\", ";
-      bool firstPage=TRUE;
-      if (quickIndex)
-      {
-        t << "[ " << endl;
-        int j;
-        for (j=33;j<127;j++)
-        {
-          if (indexLetterUsed[i][j].count()>0)
-          {
-            if (!firstPage) t << "," << endl;
-            QCString fullName = getInfo(i)->fname;
-            QCString extension = Doxygen::htmlFileExtension;
-            QCString anchor;
-            if (firstPage || !multiIndexPage) 
-              anchor=fullName+extension+"#index_";
-            else 
-              anchor=fullName+QCString().sprintf("_0x%02x",j)+extension+"#index_";
-            char is[2];is[0]=(char)j;is[1]='\0';
-            QCString ci = letterToLabel((char)j);
-            t << indentStr << "    [ ";
-            t << "\"" << is << "\", ";
-            t << "\"" << anchor << ci << "\", null ]";
-            firstPage=FALSE;
-          }
-        }
-        t << endl << indentStr << "  ] ]";
-      }
-      else
-      {
-        t << "null" << " ]";
-      }
-    }
-  }
-  return found;
-}
-
-//----------------------------------------------------------------------------
-
-#if 0
-static bool writeFullNavIndex(FTextStream &t, LayoutNavEntry *root,int indent,bool &first)
-{
-  static struct NavEntryCountMap 
-  {
-    LayoutNavEntry::Kind kind;
-    bool hasItems;
-  } navEntryCountMap[] =
-  {
-    { LayoutNavEntry::MainPage,         TRUE                                   },
-    { LayoutNavEntry::Pages,            indexedPages>0                         },
-    { LayoutNavEntry::Modules,          documentedGroups>0                     },
-    { LayoutNavEntry::Namespaces,       documentedNamespaces>0                 },
-    { LayoutNavEntry::NamespaceList,    documentedNamespaces>0                 },
-    { LayoutNavEntry::NamespaceMembers, documentedNamespaceMembers[NMHL_All]>0 },
-    { LayoutNavEntry::Classes,          annotatedClasses>0                     },
-    { LayoutNavEntry::ClassList,        annotatedClasses>0                     },
-    { LayoutNavEntry::ClassIndex,       annotatedClasses>0                     },
-    { LayoutNavEntry::ClassHierarchy,   hierarchyClasses>0                     },
-    { LayoutNavEntry::ClassMembers,     documentedClassMembers[CMHL_All]>0     },
-    { LayoutNavEntry::Files,            documentedFiles>0                      },
-    { LayoutNavEntry::FileList,         documentedFiles>0                      },
-    { LayoutNavEntry::FileGlobals,      documentedFileMembers[FMHL_All]>0      },
-    //{ LayoutNavEntry::Dirs,             documentedDirs>0                       },
-    { LayoutNavEntry::Examples,         Doxygen::exampleSDict->count()>0       }
-  };
-
-  QCString indentStr;
-  indentStr.fill(' ',indent*2);
-  bool found=FALSE;
-  if (root->children().count()>0)
-  {
-    QListIterator<LayoutNavEntry> li(root->children());
-    LayoutNavEntry *entry;
-    for (li.toFirst();(entry=li.current());++li)
-    {
-      if (navEntryCountMap[entry->kind()].hasItems && entry->visible())
-      {
-        // terminate previous entry
-        if (!first) t << "," << endl;
-        first = FALSE;
-
-        // start entry
-        if (!found)
-        {
-          t << "[" << endl;
-        }
-        found = TRUE;
-
-        bool emptySection=TRUE;
-        t << indentStr << "  [ ";
-        t << "\"" << fixSpaces(entry->title()) << "\", ";
-        t << "\"" << entry->baseFile() << Doxygen::htmlFileExtension << "\", ";
-
-        // write children (if any)
-        bool firstChild=TRUE;
-        if (entry->kind()==LayoutNavEntry::ClassMembers)
-        {
-          emptySection = !writeMemberNavIndex(t,indent+1,CMHL_Total,documentedClassMembers,g_memberIndexLetterUsed,&getCmhlInfo,firstChild);
-        }
-        else if (entry->kind()==LayoutNavEntry::NamespaceMembers)
-        {
-          emptySection = !writeMemberNavIndex(t,indent+1,NMHL_Total,documentedNamespaceMembers,g_namespaceIndexLetterUsed,&getNmhlInfo,firstChild);
-        }
-        else if (entry->kind()==LayoutNavEntry::FileGlobals)
-        {
-          emptySection = !writeMemberNavIndex(t,indent+1,FMHL_Total,documentedFileMembers,g_fileIndexLetterUsed,&getFmhlInfo,firstChild);
-        }
-        else
-        {
-          emptySection = !writeFullNavIndex(t,entry,indent+1,firstChild);
-        }
-        // end entry
-        if (emptySection) // entry without children
-          t << "null ]";
-        else // entry with children
-          t << endl << indentStr << "  ] ]";
-      }
-    }
-  }
-  return found;
-}
-#endif
-
 //----------------------------------------------------------------------------
 
 static void countRelatedPages(int &docPages,int &indexPages)

src/search.js

@@ -5,7 +5,7 @@ function convertToId(search)
   {
     var c = search.charAt(i);
     var cn = c.charCodeAt(0);
-    if (c.match(/[a-z0-9]/))
+    if (c.match(/[a-z0-9\u0080-\uFFFF]/))
     {
       result+=c;
     }
@@ -310,22 +310,20 @@ function SearchBox(name, resultsPath, inFrame, label)
     var searchValue = this.DOMSearchField().value.replace(/^ +/, "");
 
     var code = searchValue.toLowerCase().charCodeAt(0);
-    var hexCode;
-    if (code<16) 
+    var idxChar = searchValue.substr(0, 1).toLowerCase();
+    if ( 0xD800 <= code && code <= 0xDBFF && searchValue > 1) // surrogate pair
     {
-      hexCode="0"+code.toString(16);
-    }
-    else 
-    {
-      hexCode=code.toString(16);
+      idxChar = searchValue.substr(0, 2);
     }
 
     var resultsPage;
     var resultsPageWithSearch;
     var hasResultsPage;
 
-    if (indexSectionsWithContent[this.searchIndex].charAt(code) == '1')
+    var idx = indexSectionsWithContent[this.searchIndex].indexOf(idxChar);
+    if (idx!=-1)
     {
+       var hexCode=idx.toString(16);
        resultsPage = this.resultsPath + '/' + indexSectionNames[this.searchIndex] + '_' + hexCode + '.html';
        resultsPageWithSearch = resultsPage+'?'+escape(searchValue);
        hasResultsPage = true;

src/search_functions.php

@@ -358,7 +358,7 @@ function main()
   $sorted = run_query($query);
   // Now output the HTML stuff...
   // End the HTML form
-  end_form(preg_replace("/[^a-zA-Z0-9\-\_\.]/i", " ", $query ));
+  end_form(preg_replace("/[^a-zA-Z0-9\-\_\.\x80-\xFF]/i", " ", $query ));
   // report results to the user
   report_results($sorted);
   end_page();

src/search_functions_php.h

@@ -358,7 +358,7 @@
 "  $sorted = run_query($query);\n"
 "  // Now output the HTML stuff...\n"
 "  // End the HTML form\n"
-"  end_form(preg_replace(\"/[^a-zA-Z0-9\\-\\_\\.]/i\", \" \", $query ));\n"
+"  end_form(preg_replace(\"/[^a-zA-Z0-9\\-\\_\\.\\x80-\\xFF]/i\", \" \", $query ));\n"
 "  // report results to the user\n"
 "  report_results($sorted);\n"
 "  end_page();\n"

src/search_js.h

@@ -5,7 +5,7 @@
 "  {\n"
 "    var c = search.charAt(i);\n"
 "    var cn = c.charCodeAt(0);\n"
-"    if (c.match(/[a-z0-9]/))\n"
+"    if (c.match(/[a-z0-9\\u0080-\\uFFFF]/))\n"
 "    {\n"
 "      result+=c;\n"
 "    }\n"
@@ -310,22 +310,20 @@
 "    var searchValue = this.DOMSearchField().value.replace(/^ +/, \"\");\n"
 "\n"
 "    var code = searchValue.toLowerCase().charCodeAt(0);\n"
-"    var hexCode;\n"
-"    if (code<16) \n"
+"    var idxChar = searchValue.substr(0, 1).toLowerCase();\n"
+"    if ( 0xD800 <= code && code <= 0xDBFF && searchValue > 1) // surrogate pair\n"
 "    {\n"
-"      hexCode=\"0\"+code.toString(16);\n"
-"    }\n"
-"    else \n"
-"    {\n"
-"      hexCode=code.toString(16);\n"
+"      idxChar = searchValue.substr(0, 2);\n"
 "    }\n"
 "\n"
 "    var resultsPage;\n"
 "    var resultsPageWithSearch;\n"
 "    var hasResultsPage;\n"
 "\n"
-"    if (indexSectionsWithContent[this.searchIndex].charAt(code) == '1')\n"
+"    var idx = indexSectionsWithContent[this.searchIndex].indexOf(idxChar);\n"
+"    if (idx!=-1)\n"
 "    {\n"
+"       var hexCode=idx.toString(16);\n"
 "       resultsPage = this.resultsPath + '/' + indexSectionNames[this.searchIndex] + '_' + hexCode + '.html';\n"
 "       resultsPageWithSearch = resultsPage+'?'+escape(searchValue);\n"
 "       hasResultsPage = true;\n"

src/searchindex.cpp

@@ -587,8 +587,6 @@ static const char search_script[]=
 #include "search_js.h"
 ;
 
-#define MEMBER_INDEX_ENTRIES   256
-
 #define SEARCH_INDEX_ALL         0
 #define SEARCH_INDEX_CLASSES     1
 #define SEARCH_INDEX_NAMESPACES  2
@@ -606,21 +604,31 @@ static const char search_script[]=
 #define SEARCH_INDEX_PAGES      14
 #define NUM_SEARCH_INDICES      15
 
-class SearchIndexList : public SDict< QList<Definition> >
+class SearchDefinitionList : public QList<Definition>
+{
+  public:
+    SearchDefinitionList(uint letter) : m_letter(letter) {}
+    uint letter() const { return m_letter; }
+  private:
+    uint m_letter;
+};
+
+class SearchIndexList : public SDict< SearchDefinitionList >
 {
   public:
-    SearchIndexList(int size=17) : SDict< QList<Definition> >(size,FALSE) 
+    typedef Definition ElementType;
+    SearchIndexList(uint letter) : SDict<SearchDefinitionList>(17,FALSE), m_letter(letter)
     {
       setAutoDelete(TRUE);
     }
    ~SearchIndexList() {}
     void append(Definition *d)
     {
-      QList<Definition> *l = find(d->name());
+      SearchDefinitionList *l = find(d->name());
       if (l==0)
       {
-        l=new QList<Definition>;
-        SDict< QList<Definition> >::append(d->name(),l);
+        l=new SearchDefinitionList(m_letter);
+        SDict<SearchDefinitionList>::append(d->name(),l);
       }
       l->append(d);
     }
@@ -632,10 +640,13 @@ class SearchIndexList : public SDict< QList<Definition> >
       QCString n2 = md2->first()->localName();
       return qstricmp(n1.data(),n2.data());
     }
+    uint letter() const { return m_letter; }
+  private:
+    uint m_letter;
 };
 
 static void addMemberToSearchIndex(
-         SearchIndexList symbols[NUM_SEARCH_INDICES][MEMBER_INDEX_ENTRIES],
+         LetterToIndexMap<SearchIndexList> symbols[NUM_SEARCH_INDICES],
          int symbolCount[NUM_SEARCH_INDICES],
          MemberDef *md)
 {
@@ -653,58 +664,57 @@ static void addMemberToSearchIndex(
      )
   {
     QCString n = md->name();
-    uchar charCode = (uchar)n.at(0);
-    uint letter = charCode<128 ? tolower(charCode) : charCode;
     if (!n.isEmpty()) 
     {
+      uint letter = getUtf8CodeToLower(n,0);
       bool isFriendToHide = hideFriendCompounds &&
         (QCString(md->typeString())=="friend class" || 
          QCString(md->typeString())=="friend struct" ||
          QCString(md->typeString())=="friend union");
       if (!(md->isFriend() && isFriendToHide))
       {
-        symbols[SEARCH_INDEX_ALL][letter].append(md);
+        symbols[SEARCH_INDEX_ALL].append(letter,md);
         symbolCount[SEARCH_INDEX_ALL]++;
       }
       if (md->isFunction() || md->isSlot() || md->isSignal())
       {
-        symbols[SEARCH_INDEX_FUNCTIONS][letter].append(md);
+        symbols[SEARCH_INDEX_FUNCTIONS].append(letter,md);
         symbolCount[SEARCH_INDEX_FUNCTIONS]++;
       } 
       else if (md->isVariable())
       {
-        symbols[SEARCH_INDEX_VARIABLES][letter].append(md);
+        symbols[SEARCH_INDEX_VARIABLES].append(letter,md);
         symbolCount[SEARCH_INDEX_VARIABLES]++;
       }
       else if (md->isTypedef())
       {
-        symbols[SEARCH_INDEX_TYPEDEFS][letter].append(md);
+        symbols[SEARCH_INDEX_TYPEDEFS].append(letter,md);
         symbolCount[SEARCH_INDEX_TYPEDEFS]++;
       }
       else if (md->isEnumerate())
       {
-        symbols[SEARCH_INDEX_ENUMS][letter].append(md);
+        symbols[SEARCH_INDEX_ENUMS].append(letter,md);
         symbolCount[SEARCH_INDEX_ENUMS]++;
       }
       else if (md->isEnumValue())
       {
-        symbols[SEARCH_INDEX_ENUMVALUES][letter].append(md);
+        symbols[SEARCH_INDEX_ENUMVALUES].append(letter,md);
         symbolCount[SEARCH_INDEX_ENUMVALUES]++;
       }
       else if (md->isProperty())
       {
-        symbols[SEARCH_INDEX_PROPERTIES][letter].append(md);
+        symbols[SEARCH_INDEX_PROPERTIES].append(letter,md);
         symbolCount[SEARCH_INDEX_PROPERTIES]++;
       }
       else if (md->isEvent())
       {
-        symbols[SEARCH_INDEX_EVENTS][letter].append(md);
+        symbols[SEARCH_INDEX_EVENTS].append(letter,md);
         symbolCount[SEARCH_INDEX_EVENTS]++;
       }
       else if (md->isRelated() || md->isForeign() ||
                (md->isFriend() && !isFriendToHide))
       {
-        symbols[SEARCH_INDEX_RELATED][letter].append(md);
+        symbols[SEARCH_INDEX_RELATED].append(letter,md);
         symbolCount[SEARCH_INDEX_RELATED]++;
       }
     }
@@ -716,47 +726,48 @@ static void addMemberToSearchIndex(
      )
   {
     QCString n = md->name();
-    uchar charCode = (uchar)n.at(0);
-    uint letter = charCode<128 ? tolower(charCode) : charCode;
     if (!n.isEmpty()) 
     {
-      symbols[SEARCH_INDEX_ALL][letter].append(md);
+      uint letter = getUtf8CodeToLower(n,0);
+      symbols[SEARCH_INDEX_ALL].append(letter,md);
       symbolCount[SEARCH_INDEX_ALL]++;
 
       if (md->isFunction()) 
       {
-        symbols[SEARCH_INDEX_FUNCTIONS][letter].append(md);
+        symbols[SEARCH_INDEX_FUNCTIONS].append(letter,md);
         symbolCount[SEARCH_INDEX_FUNCTIONS]++;
       }
       else if (md->isVariable()) 
       {
-        symbols[SEARCH_INDEX_VARIABLES][letter].append(md);
+        symbols[SEARCH_INDEX_VARIABLES].append(letter,md);
         symbolCount[SEARCH_INDEX_VARIABLES]++;
       }
       else if (md->isTypedef())
       {
-        symbols[SEARCH_INDEX_TYPEDEFS][letter].append(md);
+        symbols[SEARCH_INDEX_TYPEDEFS].append(letter,md);
         symbolCount[SEARCH_INDEX_TYPEDEFS]++;
       }
       else if (md->isEnumerate())
       {
-        symbols[SEARCH_INDEX_ENUMS][letter].append(md);
+        symbols[SEARCH_INDEX_ENUMS].append(letter,md);
         symbolCount[SEARCH_INDEX_ENUMS]++;
       }
       else if (md->isEnumValue())
       {
-        symbols[SEARCH_INDEX_ENUMVALUES][letter].append(md);
+        symbols[SEARCH_INDEX_ENUMVALUES].append(letter,md);
         symbolCount[SEARCH_INDEX_ENUMVALUES]++;
       }
       else if (md->isDefine())
       {
-        symbols[SEARCH_INDEX_DEFINES][letter].append(md);
+        symbols[SEARCH_INDEX_DEFINES].append(letter,md);
         symbolCount[SEARCH_INDEX_DEFINES]++;
       }
     }
   }
 }
 
+// see also function convertToId() in search.js, which should match in
+// behaviour
 static QCString searchId(const QCString &s)
 {
   int c;
@@ -765,11 +776,15 @@ static QCString searchId(const QCString &s)
   for (i=0;i<s.length();i++)
   {
     c=s.at(i);
-    if ((c>='0' && c<='9') || (c>='A' && c<='Z') || (c>='a' && c<='z'))
+    if (c>0x7f || c<0) // part of multibyte character
+    {
+      result+=(char)c;
+    }
+    else if (isalnum(c)) // simply alpha numerical character
     {
       result+=(char)tolower(c);
     }
-    else
+    else // other 'unprintable' characters
     {
       char val[4];
       sprintf(val,"_%02x",(uchar)c);
@@ -780,7 +795,7 @@ static QCString searchId(const QCString &s)
 }
 
 static int g_searchIndexCount[NUM_SEARCH_INDICES];
-static SearchIndexList g_searchIndexSymbols[NUM_SEARCH_INDICES][MEMBER_INDEX_ENTRIES];
+static LetterToIndexMap<SearchIndexList> g_searchIndexSymbols[NUM_SEARCH_INDICES];
 static const char *g_searchIndexName[NUM_SEARCH_INDICES] = 
 { 
     "all",
@@ -834,12 +849,11 @@ void writeJavascriptSearchIndex()
   ClassDef *cd;
   for (;(cd=cli.current());++cli)
   {
-    uchar charCode = (uchar)cd->localName().at(0);
-    uint letter = charCode<128 ? tolower(charCode) : charCode;
+    uint letter = getUtf8CodeToLower(cd->localName(),0);
     if (cd->isLinkable() && isId(letter))
     {
-      g_searchIndexSymbols[SEARCH_INDEX_ALL][letter].append(cd);
-      g_searchIndexSymbols[SEARCH_INDEX_CLASSES][letter].append(cd);
+      g_searchIndexSymbols[SEARCH_INDEX_ALL].append(letter,cd);
+      g_searchIndexSymbols[SEARCH_INDEX_CLASSES].append(letter,cd);
       g_searchIndexCount[SEARCH_INDEX_ALL]++;
       g_searchIndexCount[SEARCH_INDEX_CLASSES]++;
     }
@@ -850,12 +864,11 @@ void writeJavascriptSearchIndex()
   NamespaceDef *nd;
   for (;(nd=nli.current());++nli)
   {
-    uchar charCode = (uchar)nd->name().at(0);
-    uint letter = charCode<128 ? tolower(charCode) : charCode;
+    uint letter = getUtf8CodeToLower(nd->name(),0);
     if (nd->isLinkable() && isId(letter))
     {
-      g_searchIndexSymbols[SEARCH_INDEX_ALL][letter].append(nd);
-      g_searchIndexSymbols[SEARCH_INDEX_NAMESPACES][letter].append(nd);
+      g_searchIndexSymbols[SEARCH_INDEX_ALL].append(letter,nd);
+      g_searchIndexSymbols[SEARCH_INDEX_NAMESPACES].append(letter,nd);
       g_searchIndexCount[SEARCH_INDEX_ALL]++;
       g_searchIndexCount[SEARCH_INDEX_NAMESPACES]++;
     }
@@ -870,12 +883,11 @@ void writeJavascriptSearchIndex()
     FileDef *fd;
     for (;(fd=fni.current());++fni)
     {
-      uchar charCode = (uchar)fd->name().at(0);
-      uint letter = charCode<128 ? tolower(charCode) : charCode;
+      uint letter = getUtf8CodeToLower(fd->name(),0);
       if (fd->isLinkable() && isId(letter))
       {
-        g_searchIndexSymbols[SEARCH_INDEX_ALL][letter].append(fd);
-        g_searchIndexSymbols[SEARCH_INDEX_FILES][letter].append(fd);
+        g_searchIndexSymbols[SEARCH_INDEX_ALL].append(letter,fd);
+        g_searchIndexSymbols[SEARCH_INDEX_FILES].append(letter,fd);
         g_searchIndexCount[SEARCH_INDEX_ALL]++;
         g_searchIndexCount[SEARCH_INDEX_FILES]++;
       }
@@ -930,8 +942,8 @@ void writeJavascriptSearchIndex()
         uint letter = charCode<128 ? tolower(charCode) : charCode;
         if (isId(letter))
         {
-          g_searchIndexSymbols[SEARCH_INDEX_ALL][letter].append(gd);
-          g_searchIndexSymbols[SEARCH_INDEX_GROUPS][letter].append(gd);
+          g_searchIndexSymbols[SEARCH_INDEX_ALL].append(letter,gd);
+          g_searchIndexSymbols[SEARCH_INDEX_GROUPS].append(letter,gd);
           g_searchIndexCount[SEARCH_INDEX_ALL]++;
           g_searchIndexCount[SEARCH_INDEX_GROUPS]++;
         }
@@ -953,8 +965,8 @@ void writeJavascriptSearchIndex()
         uint letter = charCode<128 ? tolower(charCode) : charCode;
         if (isId(letter))
         {
-          g_searchIndexSymbols[SEARCH_INDEX_ALL][letter].append(pd);
-          g_searchIndexSymbols[SEARCH_INDEX_PAGES][letter].append(pd);
+          g_searchIndexSymbols[SEARCH_INDEX_ALL].append(letter,pd);
+          g_searchIndexSymbols[SEARCH_INDEX_PAGES].append(letter,pd);
           g_searchIndexCount[SEARCH_INDEX_ALL]++;
           g_searchIndexCount[SEARCH_INDEX_PAGES]++;
         }
@@ -970,8 +982,8 @@ void writeJavascriptSearchIndex()
       uint letter = charCode<128 ? tolower(charCode) : charCode;
       if (isId(letter))
       {
-        g_searchIndexSymbols[SEARCH_INDEX_ALL][letter].append(Doxygen::mainPage);
-        g_searchIndexSymbols[SEARCH_INDEX_PAGES][letter].append(Doxygen::mainPage);
+        g_searchIndexSymbols[SEARCH_INDEX_ALL].append(letter,Doxygen::mainPage);
+        g_searchIndexSymbols[SEARCH_INDEX_PAGES].append(letter,Doxygen::mainPage);
         g_searchIndexCount[SEARCH_INDEX_ALL]++;
         g_searchIndexCount[SEARCH_INDEX_PAGES]++;
       }
@@ -979,29 +991,29 @@ void writeJavascriptSearchIndex()
   }
   
   // sort all lists
-  int i,p;
+  int i;
   for (i=0;i<NUM_SEARCH_INDICES;i++)
   {
-    for (p=0;p<MEMBER_INDEX_ENTRIES;p++)
-    {
-      if (g_searchIndexSymbols[i][p].count()>0)
+    SIntDict<SearchIndexList>::Iterator it(g_searchIndexSymbols[i]);
+    SearchIndexList *sl;
+    for (it.toFirst();(sl=it.current());++it)
     {
-        g_searchIndexSymbols[i][p].sort();
-      }
+      sl->sort();
     }
   }
 
   // write index files
   QCString searchDirName = Config_getString("HTML_OUTPUT")+"/search";
 
-  for (i=0;i<NUM_SEARCH_INDICES;i++)
-  {
-    for (p=0;p<MEMBER_INDEX_ENTRIES;p++)
+  for (i=0;i<NUM_SEARCH_INDICES;i++) // for each index
   {
-      if (g_searchIndexSymbols[i][p].count()>0)
+    SIntDict<SearchIndexList>::Iterator it(g_searchIndexSymbols[i]);
+    SearchIndexList *sl;
+    int p=0;
+    for (it.toFirst();(sl=it.current());++it,++p) // for each letter
     {
       QCString baseName;
-        baseName.sprintf("%s_%02x",g_searchIndexName[i],p);
+      baseName.sprintf("%s_%x",g_searchIndexName[i],p);
 
       QCString fileName = searchDirName + "/"+baseName+".html";
       QCString dataFileName = searchDirName + "/"+baseName+".js";
@@ -1060,8 +1072,8 @@ void writeJavascriptSearchIndex()
         ti << "[" << endl;
         bool firstEntry=TRUE;
 
-          SDict<QList<Definition> >::Iterator li(g_searchIndexSymbols[i][p]);
-          QList<Definition> *dl;
+        SDict<SearchDefinitionList>::Iterator li(*sl);
+        SearchDefinitionList *dl;
         int itemCount=0;
         for (li.toFirst();(dl=li.current());++li)
         {
@@ -1253,7 +1265,6 @@ void writeJavascriptSearchIndex()
       }
     }
   }
-  }
 
   {
     QFile f(searchDirName+"/search.js");
@@ -1275,9 +1286,12 @@ void writeJavascriptSearchIndex()
         {
           if (!first) t << "," << endl;
           t << "  " << j << ": \"";
-          for (p=0;p<MEMBER_INDEX_ENTRIES;p++)
+
+          SIntDict<SearchIndexList>::Iterator it(g_searchIndexSymbols[i]);
+          SearchIndexList *sl;
+          for (it.toFirst();(sl=it.current());++it) // for each letter
           {
-            t << (g_searchIndexSymbols[i][p].count()>0 ? "1" : "0");
+            t << QString( QChar( sl->letter() ) ).utf8();
           }
           t << "\"";
           first=FALSE;

src/sortdict.h

@@ -108,7 +108,7 @@ class SDict
      *  \param caseSensitive indicated whether the keys should be sorted
      *         in a case sensitive way.
      */
-    SDict(int size,bool caseSensitive=TRUE) : m_sizeIndex(0)
+    SDict(int size=17,bool caseSensitive=TRUE) : m_sizeIndex(0)
     {
       m_list = new SList<T>(this);
 #if AUTORESIZE
@@ -454,7 +454,7 @@ class SIntDict
      *  \param size The size of the dictionary. Should be a prime number for
      *              best distribution of elements.
      */
-    SIntDict(int size) : m_sizeIndex(0)
+    SIntDict(int size=17) : m_sizeIndex(0)
     {
       m_list = new SIntList<T>(this);
 #if AUTORESIZE
@@ -659,6 +659,76 @@ class SIntDict
         QListIterator<T> *m_li;
     };
 
+    class IteratorDict;         // first forward declare
+    friend class IteratorDict;  // then make it a friend
+    /*! Simple iterator for SDict. It iterates over the dictionary elements
+     *  in an unsorted way, but does provide information about the element's key.
+     */
+    class IteratorDict
+    {
+      public:
+        /*! Create an iterator given the dictionary. */
+        IteratorDict(const SIntDict<T> &dict)
+        {
+          m_di = new QIntDictIterator<T>(*dict.m_dict);
+        }
+
+        /*! Destroys the dictionary */
+        virtual ~IteratorDict()
+        {
+          delete m_di;
+        }
+
+        /*! Set the iterator to the first element in the list. 
+         *  \return The first compound, or zero if the list was empty. 
+         */
+        T *toFirst() const
+        {
+          return m_di->toFirst();
+        }
+
+        /*! Set the iterator to the last element in the list. 
+         *  \return The first compound, or zero if the list was empty. 
+         */
+        T *toLast() const
+        {
+          return m_di->toLast();
+        }
+
+        /*! Returns the current compound */
+        T *current() const
+        {
+          return m_di->current();
+        }
+        
+        /*! Returns the current key */
+        int currentKey() const
+        {
+          return m_di->currentKey();
+        }
+        
+        /*! Moves the iterator to the next element.
+         *  \return the new "current" element, or zero if the iterator was
+         *          already pointing at the last element.
+         */
+        T *operator++()
+        {
+          return m_di->operator++();
+        }
+
+        /*! Moves the iterator to the previous element.
+         *  \return the new "current" element, or zero if the iterator was
+         *          already pointing at the first element.
+         */
+        T *operator--()
+        {
+          return m_di->operator--();
+        }
+
+      private:
+        QDictIterator<T> *m_di;
+    };
+
 };
 
 #endif

src/util.cpp

@@ -7919,3 +7919,72 @@ void addDocCrossReference(MemberDef *src,MemberDef *dst)
   }
 }
 
+//--------------------------------------------------------------------------------------
+
+/*! @brief Get one unicode character as an unsigned integer from utf-8 string
+ *
+ * @param s utf-8 encoded string
+ * @param idx byte position of given string \a s.
+ * @return the unicode codepoint, 0 - MAX_UNICODE_CODEPOINT
+ * @see getNextUtf8OrToLower()
+ * @see getNextUtf8OrToUpper()
+ */
+uint getUtf8Code( const QCString& s, int idx )
+{
+  const int length = s.length();
+  if (idx >= length) { return 0; }
+  const uint c0 = (uchar)s.at(idx);
+  if ( c0 < 0xC2 || c0 >= 0xF8 ) // 1 byte character
+  {
+    return c0;
+  }
+  if (idx+1 >= length) { return 0; }
+  const uint c1 = ((uchar)s.at(idx+1)) & 0x3f;
+  if ( c0 < 0xE0 ) // 2 byte character
+  {
+    return ((c0 & 0x1f) << 6) | c1;
+  }
+  if (idx+2 >= length) { return 0; }
+  const uint c2 = ((uchar)s.at(idx+2)) & 0x3f;
+  if ( c0 < 0xF0 ) // 3 byte character
+  {
+    return ((c0 & 0x0f) << 12) | (c1 << 6) | c2;
+  }
+  if (idx+3 >= length) { return 0; }
+  // 4 byte character
+  const uint c3 = ((uchar)s.at(idx+3)) & 0x3f;
+  return ((c0 & 0x07) << 18) | (c1 << 12) | (c2 << 6) | c3;
+}
+
+
+/*! @brief Returns one unicode character as an unsigned integer 
+ *  from utf-8 string, making the character lower case if it was upper case.
+ *
+ * @param s utf-8 encoded string
+ * @param idx byte position of given string \a s.
+ * @return the unicode codepoint, 0 - MAX_UNICODE_CODEPOINT, excludes 'A'-'Z'
+ * @see getNextUtf8Code()
+*/
+uint getUtf8CodeToLower( const QCString& s, int idx )
+{
+  const uint v = getUtf8Code( s, idx );
+  return v < 0x7f ? tolower( v ) : v;
+}
+
+
+/*! @brief Returns one unicode character as ian unsigned interger 
+ *  from utf-8 string, making the character upper case if it was lower case.
+ *
+ * @param s utf-8 encoded string
+ * @param idx byte position of given string \a s.
+ * @return the unicode codepoint, 0 - MAX_UNICODE_CODEPOINT, excludes 'A'-'Z'
+ * @see getNextUtf8Code()
+ */
+uint getUtf8CodeToUpper( const QCString& s, int idx )
+{
+  const uint v = getUtf8Code( s, idx );
+  return v < 0x7f ? toupper( v ) : v;
+}
+
+//--------------------------------------------------------------------------------------
+

src/util.h

@@ -25,6 +25,7 @@
 #include <qlist.h>
 #include <ctype.h>
 #include "types.h"
+#include "sortdict.h"
 
 //--------------------------------------------------------------------
 
@@ -87,6 +88,33 @@ class TextGeneratorOLImpl : public TextGeneratorIntf
 
 //--------------------------------------------------------------------
 
+/** @brief maps a unicode character code to a list of T::ElementType's
+ */
+template<class T>
+class LetterToIndexMap : public SIntDict<T>
+{
+  public:
+    LetterToIndexMap() { SIntDict<T>::setAutoDelete(TRUE); }
+    int compareItems(QCollection::Item item1, QCollection::Item item2)
+    {
+      T *l1=(T *)item1;
+      T *l2=(T *)item2;
+      return (int)l1->letter()-(int)l2->letter();
+    }
+    void append(uint letter,typename T::ElementType *elem)
+    {
+      T *l = SIntDict<T>::find((int)letter);
+      if (l==0)
+      {
+        l = new T(letter);
+        SIntDict<T>::inSort((int)letter,l);
+      }
+      l->append(elem);
+    }
+};
+
+//--------------------------------------------------------------------
+
 QCString langToString(SrcLangExt lang);
 QCString getLanguageSpecificSeparator(SrcLangExt lang,bool classScope=FALSE);
 
@@ -411,5 +439,9 @@ bool fileVisibleInIndex(FileDef *fd,bool &genSourceFile);
 
 void addDocCrossReference(MemberDef *src,MemberDef *dst);
 
+uint getUtf8Code( const QCString& s, int idx );
+uint getUtf8CodeToLower( const QCString& s, int idx );
+uint getUtf8CodeToUpper( const QCString& s, int idx );
+
 #endif