TextFragment subText(const TextFragment& frag, int start, int end)
{
// this impl does an unneccesary copy, to keep TextFragment very simple for now.
if(!frag) return TextFragment();
if(start >= end) return TextFragment();
// we won't know the output fragment size in bytes until iterating the code points.
int len = frag.lengthInBytes();
SmallStackBuffer<char> temp(len);
char* buf = temp.data();
char* pb = buf;
auto first = codepoint_iterator<const char*>(frag.getText());
auto it = first;
for(int i=0; i<start; ++i)
{
++it;
}
for (int i=0; i<end - start; ++i)
{
// write the codepoint as UTF-8 to the buffer
if(!utf::internal::validate_codepoint(*it)) return TextFragment();
pb = utf::internal::utf_traits<utf::utf8>::encode(*it, pb);
++it;
}
return TextFragment(buf, pb - buf);
}
int SymbolTable::getSymbolID(const HashedCharArray& hsl)
{
int r = 0;
// get the vector of symbol IDs matching this hash. It probably has one entry but may have more.
const std::vector<int>& bin = mHashTable[hsl.hash].mIDVector;
{
bool found = false;
std::unique_lock<std::mutex> lock(mHashTable[hsl.hash].mMutex);
for(int ID : bin)
{
// there should be few collisions, so probably the first ID in the hash bin
// will be the symbol we are looking for. Unfortunately to test for equality we may have to
// compare the entire string.
TextFragment* binFragment = &mSymbolTextsByID[ID];
if(compareSizedCharArrays(binFragment->getText(), binFragment->lengthInBytes(), hsl.pChars, hsl.len))
{
r = ID;
found = true;
break;
}
}
if(!found)
{
mSymbolTextsByID.emplace_back(TextFragment(hsl.pChars, static_cast<int>(hsl.len)));
r = mSize++;
mHashTable[hsl.hash].mIDVector.emplace_back(r);
}
}
return r;
}
TextFragment::TextFragment(const TextFragment& a) noexcept
{
construct(a.getText(), a.lengthInBytes());
}
TextFragment::TextFragment(const TextFragment& t1, const TextFragment& t2, const TextFragment& t3, const TextFragment& t4) noexcept
{
construct(t1.getText(), t1.lengthInBytes(), t2.getText(), t2.lengthInBytes(),
t3.getText(), t3.lengthInBytes(), t4.getText(), t4.lengthInBytes());
}
