18#include <sdsl/bit_vectors.hpp>
35struct bin_count :
public detail::strong_type<size_t, bin_count, detail::strong_type_skill::convert>
37 using detail::strong_type<size_t,
bin_count, detail::strong_type_skill::convert>::strong_type;
42struct bin_size :
public detail::strong_type<size_t, bin_size, detail::strong_type_skill::convert>
44 using detail::strong_type<size_t,
bin_size, detail::strong_type_skill::convert>::strong_type;
49struct hash_function_count :
public detail::strong_type<size_t, hash_function_count, detail::strong_type_skill::convert>
51 using detail::strong_type<size_t,
hash_function_count, detail::strong_type_skill::convert>::strong_type;
56struct bin_index :
public detail::strong_type<size_t, bin_index, detail::strong_type_skill::convert>
58 using detail::strong_type<size_t,
bin_index, detail::strong_type_skill::convert>::strong_type;
132template <data_layout data_layout_mode_ = data_layout::uncompressed>
137 template <data_layout data_layout_mode>
148 size_t technical_bins{};
161 13043817825332782213ULL,
162 10650232656628343401ULL,
163 16499269484942379435ULL,
164 4893150838803335377ULL};
173 inline constexpr size_t hash_and_fit(
size_t h,
size_t const seed)
const
176 assert(hash_shift < 64);
177 h ^= h >> hash_shift;
178 h *= 11400714819323198485ULL;
180#ifdef __SIZEOF_INT128__
181 h =
static_cast<uint64_t
>((
static_cast<__uint128_t
>(h) *
static_cast<__uint128_t
>(bin_size_)) >> 64);
195 template <std::
integral value_t>
227 bin_size_ =
size.get();
228 hash_funs = funs.get();
232 if (hash_funs == 0 || hash_funs > 5)
233 throw std::logic_error{
"The number of hash functions must be > 0 and <= 5."};
238 bin_words = (bins + 63) >> 6;
239 technical_bins = bin_words << 6;
240 data = sdsl::bit_vector(technical_bins * bin_size_);
257 std::tie(bins, technical_bins, bin_size_, hash_shift, bin_words, hash_funs) =
258 std::tie(ibf.bins, ibf.technical_bins, ibf.bin_size_, ibf.hash_shift, ibf.bin_words, ibf.hash_funs);
260 data = sdsl::sd_vector<>{ibf.data};
282 assert(bin.get() < bins);
283 for (
size_t i = 0; i < hash_funs; ++i)
285 size_t idx = hash_and_fit(value, hash_seeds[i]);
287 assert(idx < data.size());
306 assert(bin.get() < bins);
307 for (
size_t idx = bin.get(), i = 0; i < bin_size_; idx += technical_bins, ++i)
324 template <
typename rng_t>
326 void clear(rng_t && bin_range)
noexcept
328 static_assert(std::ranges::forward_range<rng_t>,
"The range of bins to clear must model a forward_range.");
329 static_assert(std::same_as<std::remove_cvref_t<std::ranges::range_reference_t<rng_t>>,
bin_index>,
330 "The reference type of the range to clear must be seqan3::bin_index.");
332 for (
auto && bin : bin_range)
333 assert(bin.get() < bins);
336 for (
size_t offset = 0, i = 0; i < bin_size_;
offset += technical_bins, ++i)
337 for (
auto && bin : bin_range)
338 data[bin.get() +
offset] = 0;
367 size_t new_bins = new_bins_.get();
373 size_t new_bin_words = (new_bins + 63) >> 6;
377 if (new_bin_words == bin_words)
380 size_t new_technical_bins = new_bin_words << 6;
381 size_t new_bits = bin_size_ * new_technical_bins;
383 size_t idx_{new_bits}, idx{data.size()};
384 size_t delta = new_technical_bins - technical_bins + 64;
386 data.resize(new_bits);
388 for (
size_t i = idx_, j = idx; j > 0; i -= new_technical_bins, j -= technical_bins)
390 size_t stop = i - new_technical_bins;
392 for (
size_t ii = i - delta, jj = j - 64; stop && ii >= stop; ii -= 64, jj -= 64)
394 uint64_t old = data.get_int(jj);
396 data.set_int(ii, old);
400 bin_words = new_bin_words;
401 technical_bins = new_technical_bins;
435 template <
typename value_t = u
int16_t>
511 return !(lhs == rhs);
544 template <cereal_archive archive_t>
545 void CEREAL_SERIALIZE_FUNCTION_NAME(archive_t & archive)
548 archive(technical_bins);
568template <data_layout data_layout_mode>
576 ibf_t const * ibf_ptr{
nullptr};
624 assert(ibf_ptr !=
nullptr);
625 assert(result_buffer.
size() == ibf_ptr->bin_count());
628 std::memcpy(&bloom_filter_indices, &ibf_ptr->hash_seeds,
sizeof(
size_t) * ibf_ptr->hash_funs);
630 for (
size_t i = 0; i < ibf_ptr->hash_funs; ++i)
631 bloom_filter_indices[i] = ibf_ptr->hash_and_fit(value, bloom_filter_indices[i]);
633 for (
size_t batch = 0; batch < ibf_ptr->bin_words; ++batch)
636 for (
size_t i = 0; i < ibf_ptr->hash_funs; ++i)
638 assert(bloom_filter_indices[i] < ibf_ptr->
data.size());
639 tmp &= ibf_ptr->data.get_int(bloom_filter_indices[i]);
640 bloom_filter_indices[i] += 64;
643 result_buffer.data.set_int(batch << 6, tmp);
646 return result_buffer;
656template <data_layout data_layout_mode>
661 using data_type = sdsl::bit_vector;
723 return lhs.data == rhs.data;
729 return !(lhs == rhs);
763 constexpr data_type
const &
raw_data() const noexcept
793template <std::
integral value_t>
801 template <
typename binning_bitvector_t>
802 static constexpr bool is_binning_bitvector =
803 std::same_as<binning_bitvector_t,
805 || std::same_as<binning_bitvector_t,
819 using base_t::base_t;
834 template <
typename binning_bitvector_t>
835 requires is_binning_bitvector<binning_bitvector_t>
838 for_each_set_bin(binning_bitvector,
839 [
this](
size_t const bin)
853 template <
typename binning_bitvector_t>
854 requires is_binning_bitvector<binning_bitvector_t>
857 for_each_set_bin(binning_bitvector,
858 [
this](
size_t const bin)
860 assert((*
this)[bin] > 0);
908 template <
typename binning_bitvector_t,
typename on_bin_fn_t>
909 void for_each_set_bin(binning_bitvector_t && binning_bitvector, on_bin_fn_t && on_bin_fn)
911 assert(this->
size() >= binning_bitvector.size());
914 auto jump_to_next_1bit = [](
size_t & x)
922 for (
size_t bit_pos = 0; bit_pos < binning_bitvector.size(); bit_pos += 64)
925 size_t bit_sequence = binning_bitvector.raw_data().get_int(bit_pos);
928 for (
size_t bin = bit_pos; bit_sequence != 0u; ++bin, bit_sequence >>= 1)
931 bin += jump_to_next_1bit(bit_sequence);
948template <data_layout data_layout_mode>
949template <std::
integral value_t>
957 ibf_t const * ibf_ptr{
nullptr};
978 ibf_ptr(
std::addressof(ibf)),
1009 template <std::ranges::range value_range_t>
1012 assert(ibf_ptr !=
nullptr);
1013 assert(result_buffer.
size() == ibf_ptr->bin_count());
1015 static_assert(std::ranges::input_range<value_range_t>,
"The values must model input_range.");
1016 static_assert(std::unsigned_integral<std::ranges::range_value_t<value_range_t>>,
1017 "An individual value must be an unsigned integral.");
1021 for (
auto && value : values)
1024 return result_buffer;
1029 template <std::ranges::range value_range_t>
Adaptions of concepts from the Cereal library.
A data structure that behaves like a std::vector and can be used to consolidate the results of multip...
Definition: interleaved_bloom_filter.hpp:795
counting_vector & operator+=(counting_vector const &rhs)
Bin-wise addition of two seqan3::counting_vectors.
Definition: interleaved_bloom_filter.hpp:876
counting_vector & operator=(counting_vector const &)=default
Defaulted.
counting_vector & operator=(counting_vector &&)=default
Defaulted.
counting_vector & operator+=(binning_bitvector_t const &binning_bitvector)
Bin-wise adds the bits of a seqan3::interleaved_bloom_filter::membership_agent_type::binning_bitvecto...
Definition: interleaved_bloom_filter.hpp:836
counting_vector(counting_vector const &)=default
Defaulted.
~counting_vector()=default
Defaulted.
counting_vector(counting_vector &&)=default
Defaulted.
counting_vector & operator-=(binning_bitvector_t const &binning_bitvector)
Bin-wise subtracts the bits of a seqan3::interleaved_bloom_filter::membership_agent_type::binning_bit...
Definition: interleaved_bloom_filter.hpp:855
counting_vector()=default
Defaulted.
counting_vector & operator-=(counting_vector const &rhs)
Bin-wise substraction of two seqan3::counting_vectors.
Definition: interleaved_bloom_filter.hpp:889
Manages counting ranges of values for the seqan3::interleaved_bloom_filter.
Definition: interleaved_bloom_filter.hpp:951
counting_vector< value_t > const & bulk_count(value_range_t &&values) &noexcept
Counts the occurrences in each bin for all values in a range.
Definition: interleaved_bloom_filter.hpp:1010
~counting_agent_type()=default
Defaulted.
counting_agent_type()=default
Defaulted.
counting_agent_type(counting_agent_type &&)=default
Defaulted.
counting_agent_type(counting_agent_type const &)=default
Defaulted.
counting_vector< value_t > const & bulk_count(value_range_t &&values) &&noexcept=delete
Counts the occurrences in each bin for all values in a range.
counting_agent_type & operator=(counting_agent_type const &)=default
Defaulted.
counting_vector< value_t > result_buffer
Stores the result of bulk_count().
Definition: interleaved_bloom_filter.hpp:985
counting_agent_type & operator=(counting_agent_type &&)=default
Defaulted.
A bitvector representing the result of a call to bulk_contains of the seqan3::interleaved_bloom_filte...
Definition: interleaved_bloom_filter.hpp:658
binning_bitvector(binning_bitvector &&)=default
Defaulted.
constexpr data_type & raw_data() noexcept
Provides direct, unsafe access to the underlying data structure.
Definition: interleaved_bloom_filter.hpp:757
auto end() noexcept
Returns an iterator to the element following the last element of the container.
Definition: interleaved_bloom_filter.hpp:705
constexpr data_type const & raw_data() const noexcept
Provides direct, unsafe access to the underlying data structure.
Definition: interleaved_bloom_filter.hpp:763
~binning_bitvector()=default
Defaulted.
binning_bitvector(binning_bitvector const &)=default
Defaulted.
auto end() const noexcept
Returns an iterator to the element following the last element of the container.
Definition: interleaved_bloom_filter.hpp:711
auto operator[](size_t const i) const noexcept
Return the i-th element.
Definition: interleaved_bloom_filter.hpp:744
binning_bitvector(size_t const size)
Construct with given size.
Definition: interleaved_bloom_filter.hpp:679
size_t size() const noexcept
Returns the number of elements.
Definition: interleaved_bloom_filter.hpp:684
binning_bitvector & operator=(binning_bitvector &&)=default
Defaulted.
auto begin() noexcept
Returns an iterator to the first element of the container.
Definition: interleaved_bloom_filter.hpp:693
auto begin() const noexcept
Returns an iterator to the first element of the container.
Definition: interleaved_bloom_filter.hpp:699
auto operator[](size_t const i) noexcept
Return the i-th element.
Definition: interleaved_bloom_filter.hpp:737
binning_bitvector & operator=(binning_bitvector const &)=default
Defaulted.
binning_bitvector()=default
Defaulted.
friend bool operator==(binning_bitvector const &lhs, binning_bitvector const &rhs) noexcept
Test for equality.
Definition: interleaved_bloom_filter.hpp:721
friend bool operator!=(binning_bitvector const &lhs, binning_bitvector const &rhs) noexcept
Test for inequality.
Definition: interleaved_bloom_filter.hpp:727
Manages membership queries for the seqan3::interleaved_bloom_filter.
Definition: interleaved_bloom_filter.hpp:570
binning_bitvector const & bulk_contains(size_t const value) &noexcept
Determines set membership of a given value.
Definition: interleaved_bloom_filter.hpp:622
~membership_agent_type()=default
Defaulted.
membership_agent_type & operator=(membership_agent_type const &)=default
Defaulted.
binning_bitvector const & bulk_contains(size_t const value) &&noexcept=delete
Determines set membership of a given value.
membership_agent_type(membership_agent_type &&)=default
Defaulted.
membership_agent_type(membership_agent_type const &)=default
Defaulted.
membership_agent_type & operator=(membership_agent_type &&)=default
Defaulted.
membership_agent_type()=default
Defaulted.
binning_bitvector result_buffer
Stores the result of bulk_contains().
Definition: interleaved_bloom_filter.hpp:600
The IBF binning directory. A data structure that efficiently answers set-membership queries for multi...
Definition: interleaved_bloom_filter.hpp:134
interleaved_bloom_filter(interleaved_bloom_filter< data_layout::uncompressed > const &ibf)
Construct a compressed Interleaved Bloom Filter.
Definition: interleaved_bloom_filter.hpp:254
void emplace(size_t const value, bin_index const bin) noexcept
Inserts a value into a specific bin.
Definition: interleaved_bloom_filter.hpp:279
interleaved_bloom_filter & operator=(interleaved_bloom_filter const &)=default
Defaulted.
membership_agent_type membership_agent() const
Returns a seqan3::interleaved_bloom_filter::membership_agent_type to be used for lookup.
Definition: interleaved_bloom_filter.hpp:419
size_t hash_function_count() const noexcept
Returns the number of hash functions used in the Interleaved Bloom Filter.
Definition: interleaved_bloom_filter.hpp:448
constexpr data_type const & raw_data() const noexcept
Provides direct, unsafe access to the underlying data structure.
Definition: interleaved_bloom_filter.hpp:531
void clear(bin_index const bin) noexcept
Clears a specific bin.
Definition: interleaved_bloom_filter.hpp:303
interleaved_bloom_filter(seqan3::bin_count bins_, seqan3::bin_size size, seqan3::hash_function_count funs=seqan3::hash_function_count{2u})
Construct an uncompressed Interleaved Bloom Filter.
Definition: interleaved_bloom_filter.hpp:221
interleaved_bloom_filter()=default
Defaulted.
counting_agent_type< value_t > counting_agent() const
Returns a seqan3::interleaved_bloom_filter::counting_agent_type to be used for counting.
Definition: interleaved_bloom_filter.hpp:436
constexpr data_type & raw_data() noexcept
Provides direct, unsafe access to the underlying data structure.
Definition: interleaved_bloom_filter.hpp:525
interleaved_bloom_filter & operator=(interleaved_bloom_filter &&)=default
Defaulted.
friend bool operator!=(interleaved_bloom_filter const &lhs, interleaved_bloom_filter const &rhs) noexcept
Test for inequality.
Definition: interleaved_bloom_filter.hpp:509
interleaved_bloom_filter(interleaved_bloom_filter &&)=default
Defaulted.
void increase_bin_number_to(bin_count const new_bins_)
Increases the number of bins stored in the Interleaved Bloom Filter.
Definition: interleaved_bloom_filter.hpp:364
size_t bin_count() const noexcept
Returns the number of bins that the Interleaved Bloom Filter manages.
Definition: interleaved_bloom_filter.hpp:456
void clear(rng_t &&bin_range) noexcept
Clears a range of bins.
Definition: interleaved_bloom_filter.hpp:326
size_t bin_size() const noexcept
Returns the size of a single bin that the Interleaved Bloom Filter manages.
Definition: interleaved_bloom_filter.hpp:464
size_t bit_size() const noexcept
Returns the size of the underlying bitvector.
Definition: interleaved_bloom_filter.hpp:472
interleaved_bloom_filter(interleaved_bloom_filter const &)=default
Defaulted.
~interleaved_bloom_filter()=default
Defaulted.
friend bool operator==(interleaved_bloom_filter const &lhs, interleaved_bloom_filter const &rhs) noexcept
Test for equality.
Definition: interleaved_bloom_filter.hpp:486
static constexpr data_layout data_layout_mode
Indicates whether the Interleaved Bloom Filter is compressed.
Definition: interleaved_bloom_filter.hpp:191
@ offset
Sequence (seqan3::field::seq) relative start position (0-based), unsigned value.
data_layout
Determines if the Interleaved Bloom Filter is compressed.
Definition: interleaved_bloom_filter.hpp:28
@ uncompressed
The Interleaved Bloom Filter is uncompressed.
Definition: interleaved_bloom_filter.hpp:29
@ compressed
The Interleaved Bloom Filter is compressed.
Definition: interleaved_bloom_filter.hpp:30
constexpr size_t size
The size of a type pack.
Definition: traits.hpp:146
The main SeqAn3 namespace.
Definition: aligned_sequence_concept.hpp:29
SeqAn specific customisations in the standard namespace.
Provides basic data structure for strong types.
A strong type that represents the number of bins for the seqan3::interleaved_bloom_filter.
Definition: interleaved_bloom_filter.hpp:36
A strong type that represents the bin index for the seqan3::interleaved_bloom_filter.
Definition: interleaved_bloom_filter.hpp:57
A strong type that represents the number of bits for each bin in the seqan3::interleaved_bloom_filter...
Definition: interleaved_bloom_filter.hpp:43
A strong type that represents the number of hash functions for the seqan3::interleaved_bloom_filter.
Definition: interleaved_bloom_filter.hpp:50
strong_type for seed.
Definition: minimiser_hash.hpp:25