ArrayXd GoSUM::CModelVariables::expandedNTuple(int _at) const
{
int i,j,k,N=int(mvs.size()),eN=expandedSize(),exsize;
ArrayXd eX=ArrayXd::Zero(eN);
for ( i=j=0; i<N; i++,j+=exsize )
{ exsize=mvs[i].expandedSize();
if ( exsize==1 ) { eX(j)=mvs[i].sampleValue(_at); }
else { for ( k=0; k<exsize; k++ ) if (k==mvs[i].sampleValue(_at)) { eX(j+k) = 1.; break; } } }
return eX;
}
UString &UString::append(unsigned short c)
{
int thisOffset = rep->offset;
int length = size();
// possible cases:
if (length == 0) {
// this is empty - must make a new rep because we don't want to pollute the shared empty one
int newCapacity = expandedSize(1, 0);
UChar *d = static_cast<UChar *>(malloc(sizeof(UChar) * newCapacity));
d[0] = c;
release();
rep = Rep::create(d, 1);
rep->capacity = newCapacity;
} else if (!rep->baseString && rep->rc == 1) {
// this is direct and has refcount of 1 (so we can just alter it directly)
expandCapacity(thisOffset + length + 1);
UChar *d = const_cast<UChar *>(data());
d[length] = c;
rep->len = length + 1;
rep->_hash = 0;
} else if (thisOffset + length == usedCapacity()) {
// this reaches the end of the string - extend it and share
expandCapacity(thisOffset + length + 1);
UChar *d = const_cast<UChar *>(data());
d[length] = c;
Rep *newRep = Rep::create(rep, 0, length + 1);
release();
rep = newRep;
} else {
// this is shared with someone using more capacity, gotta make a whole new string
int newCapacity = expandedSize((length + 1), 0);
UChar *d = static_cast<UChar *>(malloc(sizeof(UChar) * newCapacity));
memcpy(d, data(), length * sizeof(UChar));
d[length] = c;
release();
rep = Rep::create(d, length);
rep->capacity = newCapacity;
}
return *this;
}
ArrayXd GoSUM::CModelVariables::expandNTuple(const ArrayXd &X) const
{
if ( X.size()!=mvs.size() ) throw "GoSUM::CModelVariables::expand error: wrong X size";
int i,j,k,N=int(mvs.size()),eN=expandedSize(),exsize;
ArrayXd eX=ArrayXd::Zero(eN);
for ( i=j=0; i<N; i++,j+=exsize )
{ exsize=mvs[i].expandedSize();
if ( exsize==1 ) { eX(j)=X(i); }
else { for ( k=0; k<exsize; k++ ) if (k==X(i)) { eX(j+k) = 1.; break; } } }
return eX;
}
void UString::expandCapacity(int requiredLength)
{
Rep *r = rep->baseString ? rep->baseString : rep;
if (requiredLength > r->capacity) {
int newCapacity = expandedSize(requiredLength, r->preCapacity);
r->buf = static_cast<UChar *>(realloc(r->buf, newCapacity * sizeof(UChar)));
r->capacity = newCapacity - r->preCapacity;
}
if (requiredLength > r->usedCapacity) {
r->usedCapacity = requiredLength;
}
}
UString &UString::append(const char *t)
{
int thisSize = size();
int thisOffset = rep->offset;
int tSize = strlen(t);
int length = thisSize + tSize;
// possible cases:
if (thisSize == 0) {
// this is empty
*this = t;
} else if (tSize == 0) {
// t is empty, we'll just return *this below.
} else if (!rep->baseString && rep->rc == 1) {
// this is direct and has refcount of 1 (so we can just alter it directly)
expandCapacity(thisOffset + length);
UChar *d = const_cast<UChar *>(data());
for (int i = 0; i < tSize; ++i)
d[thisSize+i] = t[i];
rep->len = length;
rep->_hash = 0;
} else if (thisOffset + thisSize == usedCapacity()) {
// this string reaches the end of the buffer - extend it
expandCapacity(thisOffset + length);
UChar *d = const_cast<UChar *>(data());
for (int i = 0; i < tSize; ++i)
d[thisSize+i] = t[i];
Rep *newRep = Rep::create(rep, 0, length);
release();
rep = newRep;
} else {
// this is shared with someone using more capacity, gotta make a whole new string
int newCapacity = expandedSize(length, 0);
UChar *d = static_cast<UChar *>(malloc(sizeof(UChar) * newCapacity));
memcpy(d, data(), thisSize * sizeof(UChar));
for (int i = 0; i < tSize; ++i)
d[thisSize+i] = t[i];
release();
rep = Rep::create(d, length);
rep->capacity = newCapacity;
}
return *this;
}
UString::UString(const UString &a, const UString &b)
{
int aSize = a.size();
int aOffset = a.rep->offset;
int bSize = b.size();
int bOffset = b.rep->offset;
int length = aSize + bSize;
// possible cases:
if (aSize == 0) {
// a is empty
attach(b.rep);
} else if (bSize == 0) {
// b is empty
attach(a.rep);
} else if (aOffset + aSize == a.usedCapacity() && 4 * aSize >= bSize &&
(-bOffset != b.usedPreCapacity() || aSize >= bSize)) {
// - a reaches the end of its buffer so it qualifies for shared append
// - also, it's at least a quarter the length of b - appending to a much shorter
// string does more harm than good
// - however, if b qualifies for prepend and is longer than a, we'd rather prepend
UString x(a);
x.expandCapacity(aOffset + length);
memcpy(const_cast<UChar *>(a.data() + aSize), b.data(), bSize * sizeof(UChar));
rep = Rep::create(a.rep, 0, length);
} else if (-bOffset == b.usedPreCapacity() && 4 * bSize >= aSize) {
// - b reaches the beginning of its buffer so it qualifies for shared prepend
// - also, it's at least a quarter the length of a - prepending to a much shorter
// string does more harm than good
UString y(b);
y.expandPreCapacity(-bOffset + aSize);
memcpy(const_cast<UChar *>(b.data() - aSize), a.data(), aSize * sizeof(UChar));
rep = Rep::create(b.rep, -aSize, length);
} else {
// a does not qualify for append, and b does not qualify for prepend, gotta make a whole new string
int newCapacity = expandedSize(length, 0);
UChar *d = static_cast<UChar *>(malloc(sizeof(UChar) * newCapacity));
memcpy(d, a.data(), aSize * sizeof(UChar));
memcpy(d + aSize, b.data(), bSize * sizeof(UChar));
rep = Rep::create(d, length);
rep->capacity = newCapacity;
}
}
void UString::expandPreCapacity(int requiredPreCap)
{
Rep *r = rep->baseString ? rep->baseString : rep;
if (requiredPreCap > r->preCapacity) {
int newCapacity = expandedSize(requiredPreCap, r->capacity);
int delta = newCapacity - r->capacity - r->preCapacity;
UChar *newBuf = static_cast<UChar *>(malloc(newCapacity * sizeof(UChar)));
memcpy(newBuf + delta, r->buf, (r->capacity + r->preCapacity) * sizeof(UChar));
free(r->buf);
r->buf = newBuf;
r->preCapacity = newCapacity - r->capacity;
}
if (requiredPreCap > r->usedPreCapacity) {
r->usedPreCapacity = requiredPreCap;
}
}
void GoSUM::CModelVariables::setNormalization()
{
int i,j,N=size(),eN=expandedSize(),exsize;
trn.setZero(eN);
scl.setOnes(eN);
for ( i=0,j=0; i<N; i++ ) {
const CModelVariable &aMV=mvs[i];
exsize=aMV.expandedSize();
if ( exsize==1 ) {
// trn(j)=aMV.minValue();
// scl(j)=aMV.maxValue()-aMV.minValue();
trn(j)=aMV.expectedValue();
scl(j)=sqrt(aMV.variance());
if ( fabs(scl(j))<TINY ) { trn(j)+=1.; scl(j)=1.; }
}
j+=exsize;
}
}