std::pair<size_t, bool>
ZipIntKeyIndex::IntVecLowerBound(fstring binkey) const {
assert(binkey.size() == sizeof(Int));
Int rawkey = unaligned_load<Int>(binkey.data());
if (rawkey < Int(m_minKey)) {
return std::make_pair(0, false);
}
auto indexData = m_index.data();
auto indexBits = m_index.uintbits();
auto indexMask = m_index.uintmask();
auto keysData = m_keys.data();
auto keysBits = m_keys.uintbits();
auto keysMask = m_keys.uintmask();
size_t key = size_t(rawkey - Int(m_minKey));
size_t hitPos = 0;
size_t hitKey = 0;
size_t i = 0, j = m_index.size();
while (i < j) {
size_t mid = (i + j) / 2;
hitPos = UintVecMin0::fast_get(indexData, indexBits, indexMask, mid);
hitKey = UintVecMin0::fast_get(keysData, keysBits, keysMask, hitPos);
if (hitKey < key)
i = mid + 1;
else
j = mid;
}
if (i < m_index.size()) {
hitPos = UintVecMin0::fast_get(indexData, indexBits, indexMask, i);
hitKey = UintVecMin0::fast_get(keysData, keysBits, keysMask, hitPos);
return std::make_pair(i, key == hitKey);
}
return std::make_pair(i, false);
}
std::pair<size_t, size_t>
ZipIntKeyIndex::IntVecEqualRange(fstring binkey) const {
assert(binkey.size() == sizeof(Int));
Int rawkey = unaligned_load<Int>(binkey.data());
if (rawkey < Int(m_minKey)) {
return std::make_pair(0, false);
}
auto indexData = m_index.data();
auto indexBits = m_index.uintbits();
auto indexMask = m_index.uintmask();
auto keysData = m_keys.data();
auto keysBits = m_keys.uintbits();
auto keysMask = m_keys.uintmask();
size_t key = size_t(rawkey - Int(m_minKey));
size_t i = 0, j = m_index.size();
size_t mid = 0;
while (i < j) {
mid = (i + j) / 2;
size_t hitPos = UintVecMin0::fast_get(indexData, indexBits, indexMask, mid);
size_t hitKey = UintVecMin0::fast_get(keysData, keysBits, keysMask, hitPos);
if (hitKey < key)
i = mid + 1;
else if (hitKey > key)
j = mid;
else
goto Found;
}
return std::make_pair(i, i);
Found:
size_t lo = i, hi = mid;
while (lo < hi) {
size_t mid2 = (lo + hi) / 2;
size_t hitPos = UintVecMin0::fast_get(indexData, indexBits, indexMask, mid2);
size_t hitKey = UintVecMin0::fast_get(keysData, keysBits, keysMask, hitPos);
if (hitKey < key) // for lower_bound
lo = mid2 + 1;
else
hi = mid2;
}
i = lo;
lo = mid + 1, hi = j;
while (lo < hi) {
size_t mid2 = (lo + hi) / 2;
size_t hitPos = UintVecMin0::fast_get(indexData, indexBits, indexMask, mid2);
size_t hitKey = UintVecMin0::fast_get(keysData, keysBits, keysMask, hitPos);
if (hitKey <= key) // for upper_bound
lo = mid2 + 1;
else
hi = mid2;
}
return std::make_pair(i, hi);
}
void operator()(int len, int idx, fstring value) {
if (strnlen(value.p, value.n) < value.size()) {
printf("%-20.*s idx=%08d bin=", len, text, idx);
for (int i = 0; i < value.n; ++i)
printf("%02X", (byte_t)value.p[i]);
printf("\n");
}
else {
printf("%-20.*s idx=%08d val=%.*s\n"
, len, text, idx, value.ilen(), value.data());
}
klen = len;
}
size_t split_by_any(fstring delims, Vec* F, size_t max_fields = ~size_t(0)) {
size_t dlen = delims.size();
if (0 == dlen) // empty delims redirect to blank delim
return split(' ', F);
if (1 == dlen)
return split(delims[0], F);
F->resize(0);
char *col = p, *end = p + n;
while (col <= end && F->size()+1 < max_fields) {
char* next = col;
while (next < end && memchr(delims.data(), *next, dlen) == NULL) ++next;
F->push_back(typename Vec::value_type(col, next));
*next = 0;
col = next + 1;
}
if (col <= end)
F->push_back(typename Vec::value_type(col, end));
return F->size();
}
size_t split(fstring delims, Vec* F, size_t max_fields = ~size_t(0)) {
size_t dlen = delims.size();
if (0 == dlen) // empty delims redirect to blank delim
return split(' ', F);
if (1 == dlen)
return split(delims[0], F);
F->resize(0);
char *col = p, *end = p + n;
while (col <= end && F->size()+1 < max_fields) {
char* next = (char*)memmem(col, end-col, delims.data(), dlen);
if (NULL == next) next = end;
F->push_back(typename Vec::value_type(col, next));
*next = 0;
col = next + dlen;
}
if (col <= end)
F->push_back(typename Vec::value_type(col, end));
return F->size();
}