/*!
* \param text_buf Byte stream.
* \param len Length of the byte stream.
*/
succinct_byte_alphabet(int_vector_buffer<8>& text_buf, int_vector_size_type len):
char2comp(this), comp2char(this), C(m_C), sigma(m_sigma)
{
m_sigma = 0;
if (0 == len or 0 == text_buf.size())
return;
assert(len <= text_buf.size());
// initialize vectors
int_vector<64> D(257, 0);
bit_vector tmp_char(256, 0);
// count occurrences of each symbol
for (size_type i=0; i < len; ++i) {
++D[text_buf[i]];
}
assert(1 == D[0]); // null-byte should occur exactly once
m_sigma = 0;
for (int i=0; i<256; ++i)
if (D[i]) {
tmp_char[i] = 1; // mark occurring character
D[m_sigma] = D[i]; // compactify m_C
++m_sigma;
}
// resize to sigma+1, since CSAs also need the sum of all elements
m_C = C_type(m_sigma+1, 0, bits::hi(len)+1);
for (int i=(int)m_sigma; i > 0; --i) m_C[i] = D[i-1];
m_C[0] = 0;
for (int i=1; i <= (int)m_sigma; ++i) m_C[i] = m_C[i] + m_C[i-1];
assert(m_C[sigma]==len);
m_char = tmp_char;
util::init_support(m_char_rank, &m_char);
util::init_support(m_char_select, &m_char);
}
void vcf_file::read_line(vector<char> &out)
{
static string tmp;
tmp="";
out.resize(0);
read_line(tmp);
vector<char> tmp_char(tmp.begin(),tmp.end());
out = tmp_char;
}
/*!
* \param text_buf Byte stream.
* \param len Length of the byte stream.
*/
int_alphabet_strategy(int_vector_file_buffer<0> &text_buf, int_vector_size_type len):
char2comp(this), comp2char(this), C(m_C), sigma(m_sigma)
{
m_sigma = 0;
text_buf.reset();
if (0 == len or 0 == text_buf.int_vector_size)
return;
assert( len <= text_buf.int_vector_size );
// initialize vectors
std::map<size_type, size_type> D;
// count occurrences of each symbol
for (size_type i=0, r_sum=0, r = text_buf.load_next_block(); i < len;) {
for (; i < r_sum+r; ++i) {
D[text_buf[i-r_sum]]++;
}
r_sum += r; r = text_buf.load_next_block();
}
m_sigma = D.size();
if ( is_continuous_alphabet(D) ){
// do not initialize m_char, m_char_rank and m_char_select since we can map directly
}else{
// note: the alphabet has at least size 1, so the following is safe:
size_type largest_symbol = (--D.end())->first;
bit_vector tmp_char(largest_symbol+1, 0);
for (std::map<size_type, size_type>::const_iterator it = D.begin(), end=D.end(); it != end; ++it){
tmp_char[it->first] = 1;
}
util::assign(m_char, tmp_char);
util::init_support(m_char_rank, &m_char);
util::init_support(m_char_select, &m_char);
}
assert(D.find(0) != D.end() and 1 == D[0]); // null-byte should occur exactly once
// resize to sigma+1, since CSAs also need the sum of all elements
util::assign(m_C, C_type(m_sigma+1, 0, bit_magic::l1BP(len)+1) );
size_type sum = 0, idx=0;
for (std::map<size_type, size_type>::const_iterator it = D.begin(), end=D.end(); it != end; ++it){
m_C[idx++] = sum;
sum += it->second;
}
m_C[idx] = sum; // insert sum of all elements
}
/*!
* \param text_buf Byte stream.
* \param len Length of the byte stream.
*/
succinct_byte_alphabet_strategy(int_vector_file_buffer<8> &text_buf, int_vector_size_type len):
char2comp(this), comp2char(this), C(m_C), sigma(m_sigma)
{
m_sigma = 0;
text_buf.reset();
if (0 == len or 0 == text_buf.int_vector_size)
return;
assert( len <= text_buf.int_vector_size );
// initialize vectors
int_vector<64> D(257, 0);
bit_vector tmp_char(256, 0);
// count occurrences of each symbol
for (size_type i=0, r_sum=0, r = text_buf.load_next_block(); i < len;) {
for (; i < r_sum+r; ++i) {
++D[text_buf[i-r_sum]];
}
r_sum += r; r = text_buf.load_next_block();
}
assert(1 == D[0]); // null-byte should occur exactly once
m_sigma = 0;
for (int i=0; i<256; ++i)
if (D[i]) {
tmp_char[i] = 1; // mark occurring character
D[m_sigma] = D[i]; // compactify m_C
++m_sigma;
}
// resize to sigma+1, since CSAs also need the sum of all elements
util::assign(m_C, C_type(m_sigma+1, 0, bit_magic::l1BP(len)+1));
for (int i=(int)m_sigma; i > 0; --i) m_C[i] = D[i-1];
m_C[0] = 0;
for (int i=1; i <= (int)m_sigma; ++i) m_C[i] = m_C[i] + m_C[i-1];
assert(m_C[sigma]==len);
util::assign(m_char, tmp_char);
util::init_support(m_char_rank, &m_char);
util::init_support(m_char_select, &m_char);
}