PrefixCollection.cpp

   1 // Implementation of CPrefixCollection methods
   2 // Copyright © 2009 The University of Chicago
   3 #include "PrefixCollection.h"
   4
   5 #include <Q3TextStream>
   6 #include <QIODevice>
   7 #include <QFile>
   8 #include "ui/Status.h"
   9 #include "Lexicon.h"
  10 #include "MiniLexicon.h"
  11 #include "Prefix.h"
  12
  13 CPrefixCollection::CPrefixCollection (CMiniLexicon* Lex)
  14 {
  15         m_pMiniLex = Lex;
  16         if( m_pMiniLex ) m_pLexicon = m_pMiniLex->GetLexicon();
  17 }
  18
  19
  20
  21 CPrefixCollection::CPrefixCollection ( )
  22 {
  23   m_pLexicon = NULL;
  24   m_pMiniLex = NULL;
  25 }
  26
  27
  28 CPrefixCollection::~CPrefixCollection()
  29 {
  30         if( m_pLexicon )
  31         {
  32                 CPrefix* pPrefix;
  33
  34                 for( int i = 0; i < GetCount(); i++ )
  35                 {
  36                         pPrefix = GetAt(i);
  37
  38                         m_pLexicon->RemovePrefix( pPrefix );
  39                 }
  40         }
  41 }
  42
  43
  44
  45
  46
  47 CPrefix* CPrefixCollection::operator<< ( CStringSurrogate Prefix )
  48 {
  49   CPrefix*  pPrefix;
  50
  51   pPrefix = AddToCollection( Prefix );
  52
  53   return pPrefix;
  54 }
  55
  56
  57 CPrefix* CPrefixCollection::operator<< ( CParse* Prefix )
  58 {
  59   CPrefix*  pPrefix;
  60
  61   pPrefix = AddToCollection (*Prefix);
  62
  63   return pPrefix;
  64 }
  65
  66
  67 CPrefix* CPrefixCollection::operator<< ( QString Prefix )
  68 {
  69   CPrefix*  pPrefix;
  70
  71   CStringSurrogate ssPrefix( Prefix );
  72   pPrefix = AddToCollection( ssPrefix );
  73
  74   return pPrefix;
  75 }
  76
  77
  78 void CPrefixCollection::AddPointer( CPrefix* pPrefix )
  79 {
  80         TCollection<CPrefix>::AddPointer( pPrefix );
  81
  82         if( m_pLexicon ) m_pLexicon->InsertPrefix( pPrefix );
  83 }
  84
  85
  86 CPrefix* CPrefixCollection::AddToCollection( CParse& Prefix )
  87 {
  88         CPrefix* pPrefix = TCollection<CPrefix>::AddToCollection( Prefix );
  89
  90         if( m_pLexicon ) m_pLexicon->InsertPrefix( pPrefix );
  91
  92         return pPrefix;
  93 }
  94
  95
  96 CPrefix* CPrefixCollection::AddToCollection( CStringSurrogate& Prefix )
  97 {
  98         CPrefix* pPrefix = TCollection<CPrefix>::AddToCollection( Prefix );
  99
 100         if( m_pLexicon ) m_pLexicon->InsertPrefix( pPrefix );
 101
 102         return pPrefix;
 103 }
 104
 105
 106 void CPrefixCollection::Empty()
 107 {
 108         if( m_pLexicon )
 109         {
 110                 CPrefix* pPrefix;
 111
 112                 for( int i = 0; i < GetCount(); i++ )
 113                 {
 114                         pPrefix = GetAt(i);
 115
 116                         m_pLexicon->RemovePrefix( pPrefix );
 117                 }
 118         }
 119
 120         TCollection<CPrefix>::Empty();
 121 }
 122
 123
 124 void CPrefixCollection::RemoveAll()
 125 {
 126         if( m_pLexicon )
 127         {
 128                 CPrefix* pPrefix;
 129
 130                 for( int i = 0; i < GetCount(); i++ )
 131                 {
 132                         pPrefix = GetAt(i);
 133
 134                         m_pLexicon->RemovePrefix( pPrefix );
 135                 }
 136         }
 137
 138         TCollection<CPrefix>::RemoveAll();
 139 }
 140
 141
 142 bool CPrefixCollection::Remove( CPrefix* pPrefix )
 143 {
 144         if( m_pLexicon ) m_pLexicon->RemovePrefix( pPrefix );
 145
 146         return TCollection<CPrefix>::Remove( pPrefix );
 147 }
 148
 149
 150 bool CPrefixCollection::RemoveMember( CPrefix* pPrefix )
 151 {
 152         if( m_pLexicon ) m_pLexicon->RemovePrefix( pPrefix );
 153
 154         return TCollection<CPrefix>::RemoveMember( pPrefix );
 155 }
 156
 157
 158 bool CPrefixCollection::RemoveMember( CStringSurrogate& Prefix )
 159 {
 160         CPrefix* pPrefix = (CPrefix*)Find1( Prefix )->Get_T_Pointer();
 161
 162         if( m_pLexicon ) m_pLexicon->RemovePrefix( pPrefix );
 163
 164         return TCollection<CPrefix>::RemoveMember( Prefix );
 165 }
 166
 167
 168 bool CPrefixCollection::RemoveMember( CStringSurrogate& Prefix, bool b )
 169 {
 170         CPrefix* pPrefix = (CPrefix*)Find1( Prefix )->Get_T_Pointer();
 171
 172         if( m_pLexicon ) m_pLexicon->RemovePrefix( pPrefix );
 173
 174         return TCollection<CPrefix>::RemoveMember( Prefix, b );
 175 }
 176
 177
 178 void CPrefixCollection::DeleteMarkedMembers()
 179 {
 180         if ( m_DeletionArray == NULL ) { return; }
 181
 182         int count = GetCount();
 183         for (int i = 0; i < count; i++)
 184         {
 185                 if ( m_DeletionArray[i] == 1 )
 186                 {
 187                         if( m_pLexicon ) m_pLexicon->RemovePrefix( m_PointerArray[i] );
 188                 }
 189         }
 190
 191         TCollection<CPrefix>::DeleteMarkedMembers();
 192 }
 193
 194 void CPrefixCollection::ListDisplay(
 195         Q3ListView* pView, QMap<QString, QString>* filter,
 196         linguistica::ui::status_user_agent& status)
 197 {
 198         pView->setRootIsDecorated(false);
 199         pView->setSorting(1);
 200
 201         // Clean it out first.
 202         while (pView->columns() != 0)
 203                 pView->removeColumn(0);
 204
 205         // Add Column headers
 206         pView->addColumn("Prefixes");
 207         pView->addColumn("Descr. Length");
 208         // XXX. suffixes has "Length Of Ptr" column
 209         pView->addColumn("Corpus Count");
 210         pView->addColumn("Use Count");
 211         pView->addColumn("Stems");
 212
 213         // XXX. suffixes sets column alignments and widths
 214
 215         status.major_operation = "Creating prefix list for display";
 216         status.progress.clear();
 217
 218         if (m_SortValidFlag == false)
 219                 Sort(COUNT);
 220
 221         // Display each item
 222         status.progress.set_denominator(GetCount());
 223         for (int i = 0; i < (int) GetCount(); i++) {
 224                 GetAtSort(i)->ListDisplay(pView, filter, m_pLexicon->GetNumberOfCharacterTypes());
 225                 status.progress = i;
 226         }
 227         status.progress.clear();
 228         status.major_operation.clear();
 229 }
 230
 231 void CPrefixCollection::OutputPrefixes( QString FileName )
 232 {
 233         Q_ASSERT(!FileName.isEmpty());
 234         QFile file( FileName );
 235
 236         if( file.open( QIODevice::WriteOnly ) )
 237         {
 238                 Q3TextStream outf( &file );
 239                 outf.setEncoding( Q3TextStream::Unicode );
 240
 241                 outf << "# Prefix Count" << endl;
 242                 outf << "# ------------" << endl;
 243                 outf << "  " << GetCount() << endl << endl;
 244
 245                 outf << "# Prefix       | Use Count | Corpus Count | Index | " << endl;
 246                 outf << "# ------------------------------------------------- " << endl;
 247
 248                 Sort( CORPUSCOUNT );
 249                 for (int i = 0; i < (int)GetCount(); i++)
 250                 {
 251                         GetAtSort(i)->OutputPrefix( outf );
 252                 }
 253
 254                 file.close();
 255         }
 256 }
 257
 258
 259 void  CPrefixCollection::ReadPrefixFile (QString FileName)
 260 {
 261         Q_ASSERT(!FileName.isEmpty());
 262         CPrefix*        pPrefix;
 263         QFile           file(FileName);
 264         QString Buffer, Key;
 265         int             Index,
 266         size,
 267         UseCount,
 268         CorpusCount,
 269         LineCount = 0;
 270
 271         if( file.exists() && file.open( QIODevice::ReadOnly ) )
 272         {
 273                 Q3TextStream inf(&file);
 274                 inf.setEncoding ( Q3TextStream::Locale );
 275
 276                 Buffer = inf.readLine();
 277                 Q_ASSERT( Buffer[0] == '#' );
 278
 279                 Buffer = inf.readLine();
 280                 Q_ASSERT( Buffer[0] == '#' );
 281
 282                 inf >> size;
 283
 284                 Buffer = inf.readLine(); // end of size line
 285                 Q_ASSERT( Buffer.length() == 0 );
 286
 287                 Buffer = inf.readLine(); // blank line
 288                 Q_ASSERT( Buffer.length() == 0 );
 289
 290                 Buffer = inf.readLine();
 291                 Q_ASSERT( Buffer[0] == '#' );
 292
 293                 Buffer = inf.readLine();
 294                 Q_ASSERT( Buffer[0] == '#' );
 295
 296                 while( !inf.atEnd() && LineCount < size )
 297                 {
 298                         LineCount++;
 299
 300                         inf >> Key;
 301                         inf >> UseCount;
 302                         inf >> CorpusCount;
 303                         inf >> Index;
 304
 305                         // Filter all sequences that should be
 306                         // analyzed as one character
 307                         Key = Filter( m_pLexicon->GetInFilter(), Key );
 308
 309                         pPrefix = *this << CStringSurrogate( Key );
 310                         pPrefix->IncrementUseCount( UseCount );
 311                         pPrefix->IncrementCorpusCount( CorpusCount );
 312                 }
 313
 314                 file.close();
 315         }
 316 }
 317
 318
 319 double  CPrefixCollection::GetDL_PhonologicalContent()
 320 {
 321         double affix_total_dl = 0;
 322         int i;
 323         CPrefix* pAffix;
 324         for( i = 0; i < (int)GetCount(); i++ )
 325         {
 326                 pAffix =  GetAt(i);
 327                 affix_total_dl += pAffix->ComputeDL( m_pMiniLex->GetNumberOfCharacterTypes() );
 328         }
 329         return affix_total_dl ;
 330 }