bool readReference(IBitStream& bstr, unsigned& reference, unsigned huffmanNumber) {
int code = bstr.read(2);
if (code == 3) {
reference = bstr.read(32);
return true;
}
int bitlen = BitLength(huffmanNumber);
assert(bitlen >= 2);
reference = (code << (bitlen - 2)) | bstr.read(bitlen - 2);
return true;
}
bool UserDictionaryDecoder::DecodeArticle(
IBitStream *bstr,
std::u16string &res,
std::u16string const& prefix,
LenTable& ltArticles,
std::vector<char32_t>& articleSymbols)
{
unsigned len = bstr->read(16);
if (len == 0xFFFF) {
len = bstr->read(32);
}
res.clear();
unsigned symIdx;
std::vector<uint32_t> vec;
while ((unsigned)res.length() < len) {
ltArticles.Decode(*bstr, symIdx);
unsigned sym = articleSymbols.at(symIdx);
vec.push_back(sym);
if (sym >= 0x10000) {
if (sym >= 0x10040) {
unsigned startIdx = bstr->read(BitLength(len));
unsigned len = sym - 0x1003d;
res += res.substr(startIdx, len);
vec.push_back(startIdx);
} else {
unsigned startIdx = bstr->read(BitLength(prefix.length()));
unsigned len = sym - 0xfffd;
res += prefix.substr(startIdx, len);
vec.push_back(startIdx);
}
} else {
res += (char16_t)sym;
}
}
return true;
}
void LenTable::Read(IBitStream &bitstr) {
symidx2nodeidx.clear();
nodes.clear();
int count = bitstr.read(32);
int bitsPerLen = bitstr.read(8);
int idxBitSize = BitLength(count);
symidx2nodeidx.resize(count);
for (unsigned& nodeIdx : symidx2nodeidx) {
nodeIdx = -1; // in case the root has a leaf as a child
}
nodes.resize(count - 1);
int rootIdx = nodes.size() - 1;
nodes.at(rootIdx) = {0,0,-1,-1};
nextNodePosition = 0;
for (int i = 0; i < count; ++i) {
int symidx = bitstr.read(idxBitSize);
int len = bitstr.read(bitsPerLen);
placeSymidx(symidx, rootIdx, len);
}
}
