/*
* Constructor #3
*
* We know character probabilities. Input: pairs <character, probability>
*
* Here the alphabet is Huffman encoded.
*
* Note: all characters that will appear in the text must be included
* in P. If in doubt, assign probability 0 (such characters will get
* the longest codes)
*
*/
alphabet_encoder(vector<pair<char_type,double> >& P){
sigma = P.size();
enc_type = huffman;
auto comp = [](node x, node y){ return x.second < y.second; };
multiset<node,decltype(comp)> s(comp);
//insert leaves
for(auto it = P.begin();it!=P.end();++it)
s.insert({{NULL,&it->first},it->second});
//Huffman algorithm
while(s.size()>1){
//extract, copy and erase the 2 smallest elements
auto min1 = new node(*s.begin());
s.erase(s.begin());
auto min2 = new node(*s.begin());
s.erase(s.begin());
double new_prob = min1->second + min2->second;
pair<void*,void*> children = {min1,min2};
s.insert({ children, new_prob });
}
node root = *s.begin();
extract_codes(&root,{});
root.free_memory();
}
bool NerveTool::extract_fiberitems( FILE *sout , StringList& fibersinfo , String type , String& value ) {
// parse value: x,y -> x,y -> x,y ...
ClassList<StringList> chain;
value.trim();
while( !value.isEmpty() ) {
String part;
int idx = value.find( "->" );
if( idx < 0 ) {
if( chain.count() == 0 )
return( false );
part = value;
value.clear();
}
else {
part = value.getMid( 0 , idx );
value.remove( 0 , idx + 2 );
value.trim();
if( value.isEmpty() )
return( false );
}
// parse part
StringList *z = new StringList;
chain.add( z );
if( !extract_codes( part , z ) ) {
fprintf( sout , "wrong part=%s\n" , ( const char * )part );
return( false );
}
// prohibit many-to-many
if( z -> count() > 1 && chain.count() > 1 ) {
StringList& zp = chain.getRef( chain.count() - 2 );
if( zp.count() > 1 )
return( false );
}
}
// chain of more than one
if( chain.count() < 2 )
return( false );
// split chain
int startChain = 0;
int startChainCount = 0;
for( int k = 0; k < chain.count(); k++ ) {
StringList& z = chain.getRef( k );
int zn = z.count();
// starter
if( k == 0 ) {
startChainCount = zn;
continue;
}
// many to one - split
if( startChainCount > 1 ) {
if( zn != 1 )
return( false );
addManyToOne( fibersinfo , type , chain.getRef( startChainCount ) , z.get( 0 ) );
startChain = k;
startChainCount = zn;
continue;
}
// allow x -> y -> z as is
if( zn == 1 ) {
if( k == chain.count() - 1 ) {
addSingleChain( fibersinfo , type , chain , startChain , k );
break;
}
continue;
}
// x -> y -> x,y - split to x -> y and y -> x,y
if( ( k - 1 ) > startChain ) {
addSingleChain( fibersinfo , type , chain , startChain , k - 1 );
startChain = k - 1;
startChainCount = 1;
}
addOneToMany( fibersinfo , type , chain.getRef( startChain ).get( 0 ) , z );
startChain = k;
startChainCount = zn;
}
chain.destroy();
return( true );
}
