Refactor Bitvector to use more free functions

lib/genesis/population/format/bed_reader.cpp

@@ -75,7 +75,7 @@ GenomeLocusSet BedReader::read_as_genome_locus_set(
     auto const chr_names = result.chromosome_names();
     for( auto const& chr_name : chr_names ) {
         auto& bv = result.chromosome_positions( chr_name );
-        auto const last_bit_idx = bv.find_last_set();
+        auto const last_bit_idx = find_last_set( bv );
         if( last_bit_idx == Bitvector::npos ) {
             bv = Bitvector( 1 );
         } else {
@@ -113,7 +113,7 @@ GenomeLocusSet BedReader::read_as_genome_locus_set(
 
         // Use the seq dict to resize the bitvector to the desired length.
         auto& bv = result.chromosome_positions( chr_name );
-        auto const last_bit_idx = bv.find_last_set();
+        auto const last_bit_idx = find_last_set( bv );
         if( last_bit_idx == Bitvector::npos ) {
             // Empty chr in bed. Should not really be able to happen, as that means there was not
             // an entry in the input to begin with, but let's catch it anyway.

lib/genesis/population/function/genome_locus_set.cpp

@@ -109,7 +109,7 @@ GenomeLocusSet read_mask_fasta(
         // If we invert, we do that here at the end. We could also switch in the set function
         // above, but it's easier to do this in bulk. We need to unset the first bit then.
         if( invert ) {
-            bv.negate();
+            negate(bv);
             bv.unset(0);
         }
 

lib/genesis/population/function/window_average.hpp

@@ -40,6 +40,9 @@
 #include "genesis/population/variant.hpp"
 #include "genesis/population/window/base_window.hpp"
 #include "genesis/population/window/window_view.hpp"
+#include "genesis/utils/math/bitvector.hpp"
+#include "genesis/utils/math/bitvector/functions.hpp"
+#include "genesis/utils/math/bitvector/operators.hpp"
 
 #include <cassert>
 #include <limits>
@@ -217,7 +220,7 @@ inline size_t get_window_provided_loci_count(
 
         // Finally, we have checked everything. Our first and last position are both inclusive,
         // while the bitvector count uses past-the-end, so we need to add one here for the last.
-        return bv.count( first, last + 1 );
+        return pop_count( bv, first, last + 1 );
     };
 
     // If the window is a WindowStream over a whole genome, we use all its chromosomes.

lib/genesis/population/genome_locus_set.cpp

@@ -30,6 +30,7 @@
 
 #include "genesis/population/genome_locus_set.hpp"
 
+#include "genesis/utils/math/bitvector/functions.hpp"
 #include "genesis/utils/math/bitvector/operators.hpp"
 
 #include <cassert>
@@ -90,7 +91,7 @@ void GenomeLocusSet::add(
     assert( bv.size() >= end + 1 );
 
     // Now set all bits in between the two positions, inclusive.
-    bv.set( start, end + 1 );
+    bv.set_range( start, end + 1 );
     // for( size_t i = start; i <= end; ++i ) {
     //     bv.set( i );
     // }
@@ -203,7 +204,7 @@ void GenomeLocusSet::set_intersect( GenomeLocusSet const& rhs )
         // so remove it from the to-delete list. If all its bits are 0, we have eliminated
         // all positions from the filter, so we might as well delete the whole vector; in that
         // case, we simply keept it in the to-delete list and then it gets removed below.
-        if( lhs_bits.count() > 0 ) {
+        if( pop_count(lhs_bits) > 0 ) {
             chrs_to_delete.erase( chr.first );
         }
     }

lib/genesis/population/genome_locus_set.hpp

@@ -44,6 +44,7 @@
 #include "genesis/population/genome_region_list.hpp"
 #include "genesis/sequence/sequence_dict.hpp"
 #include "genesis/utils/math/bitvector.hpp"
+#include "genesis/utils/math/bitvector/functions.hpp"
 
 namespace genesis {
 namespace population {
@@ -386,7 +387,7 @@ class GenomeLocusSet
 
         // We do not need to to an extra check for position 0 here.
         // If it is true, then so will be the result.
-        return bv.any_set();
+        return any_set( bv );
     }
 
     // -------------------------------------------------------------------------
@@ -422,7 +423,7 @@ class GenomeLocusSet
 
         // If the above is not the case, check the actual start_position.
         // If the start_position is outside of the bitvector, it is not covered, obviously.
-        return bitvector.find_next_set( start_position );
+        return find_next_set( bitvector, start_position );
     }
 
     /**

lib/genesis/sequence/functions/functions.cpp

@@ -1,6 +1,6 @@
 /*
     Genesis - A toolkit for working with phylogenetic data.
-    Copyright (C) 2014-2019 Lucas Czech and HITS gGmbH
+    Copyright (C) 2014-2024 Lucas Czech
 
     This program is free software: you can redistribute it and/or modify
     it under the terms of the GNU General Public License as published by
@@ -35,6 +35,9 @@
 #include "genesis/sequence/printers/simple.hpp"
 
 #include "genesis/utils/core/logging.hpp"
+#include "genesis/utils/math/bitvector.hpp"
+#include "genesis/utils/math/bitvector/functions.hpp"
+#include "genesis/utils/math/bitvector/operators.hpp"
 #include "genesis/utils/text/string.hpp"
 #include "genesis/utils/text/style.hpp"
 
@@ -189,7 +192,7 @@ void remove_sites( Sequence& seq, utils::Bitvector sites )
         );
     }
 
-    auto const num_sites = sites.size() - sites.count();
+    auto const num_sites = sites.size() - pop_count(sites);
     std::string result;
     result.reserve( num_sites );
 

lib/genesis/tree/bipartition/bipartition.hpp

@@ -3,7 +3,7 @@
 
 /*
     Genesis - A toolkit for working with phylogenetic data.
-    Copyright (C) 2014-2019 Lucas Czech and HITS gGmbH
+    Copyright (C) 2014-2024 Lucas Czech
 
     This program is free software: you can redistribute it and/or modify
     it under the terms of the GNU General Public License as published by
@@ -34,6 +34,7 @@
 #include "genesis/tree/tree.hpp"
 #include "genesis/tree/tree/subtree.hpp"
 #include "genesis/utils/math/bitvector.hpp"
+#include "genesis/utils/math/bitvector/functions.hpp"
 
 #include <stdexcept>
 
@@ -96,7 +97,7 @@ class Bipartition
 
     void invert()
     {
-        leaf_nodes_.negate();
+        negate(leaf_nodes_);
         link_ = &link_->outer();
     }
 

lib/genesis/tree/bipartition/functions.cpp

@@ -1,6 +1,6 @@
 /*
     Genesis - A toolkit for working with phylogenetic data.
-    Copyright (C) 2014-2020 Lucas Czech and HITS gGmbH
+    Copyright (C) 2014-2024 Lucas Czech
 
     This program is free software: you can redistribute it and/or modify
     it under the terms of the GNU General Public License as published by
@@ -249,18 +249,18 @@ Bipartition find_smallest_subtree(
 
         auto const inverted = ~(bip.leaf_nodes());
         if( utils::is_subset( comp, bip.leaf_nodes() )) {
-            if( min_count == 0 || bip.leaf_nodes().count() < min_count ) {
+            if( min_count == 0 || pop_count(bip.leaf_nodes()) < min_count ) {
                 best_bip = bip;
-                min_count = best_bip.leaf_nodes().count();
+                min_count = pop_count(best_bip.leaf_nodes());
             }
         }
         if( utils::is_subset( comp, inverted ) ) {
-            if( min_count == 0 || inverted.count() < min_count ) {
+            if( min_count == 0 || pop_count(inverted) < min_count ) {
                 best_bip = bip;
                 best_bip.invert();
 
-                assert( best_bip.leaf_nodes().count() == inverted.count() );
-                min_count = best_bip.leaf_nodes().count();
+                assert( pop_count( best_bip.leaf_nodes() ) == pop_count( inverted ) );
+                min_count = pop_count( best_bip.leaf_nodes() );
             }
         }
     }

lib/genesis/tree/bipartition/rf.cpp

@@ -1,6 +1,6 @@
 /*
     Genesis - A toolkit for working with phylogenetic data.
-    Copyright (C) 2014-2019 Lucas Czech
+    Copyright (C) 2014-2024 Lucas Czech
 
     This program is free software: you can redistribute it and/or modify
     it under the terms of the GNU General Public License as published by
@@ -34,6 +34,8 @@
 #include "genesis/tree/function/functions.hpp"
 #include "genesis/tree/iterator/postorder.hpp"
 #include "genesis/utils/core/algorithm.hpp"
+#include "genesis/utils/math/bitvector/functions.hpp"
+#include "genesis/utils/math/bitvector/operators.hpp"
 
 #include <cstdint>
 #include <algorithm>
@@ -201,14 +203,14 @@ void rf_get_bitvectors_template(
 
             // Call the bitvector processor functor now, as we just finished constructing a split.
             // We normalize first to make sure that we always get comparable bitvectors in the end.
-            current.normalize();
+            normalize(current);
             process_bitvector( current );
         }
     }
 
     // We have traversed all node names now. If there is still an unset bit in the bitvector,
     // that means that we did not find all names that are in the tree.
-    if( name_check.count() != names.size() ) {
+    if( pop_count(name_check) != names.size() ) {
         throw std::runtime_error(
             "Cannot calculate RF distance with trees that have different node names. "
             "Some names are missing from one of the trees."

lib/genesis/utils.hpp

@@ -139,6 +139,7 @@
 #include "genesis/utils/io/string_output_target.hpp"
 #include "genesis/utils/math/binomial.hpp"
 #include "genesis/utils/math/bit.hpp"
+#include "genesis/utils/math/bitvector/functions.hpp"
 #include "genesis/utils/math/bitvector.hpp"
 #include "genesis/utils/math/bitvector/operators.hpp"
 #include "genesis/utils/math/common.hpp"

lib/genesis/utils/math/bit.hpp

@@ -36,7 +36,9 @@
 #include "genesis/utils/core/std.hpp"
 
 #include <array>
+#include <climits>
 #include <cstdint>
+#include <string>
 #include <type_traits>
 
 #if GENESIS_CPP_STD >= GENESIS_CPP_STD_20
@@ -53,6 +55,33 @@
 namespace genesis {
 namespace utils {
 
+// ================================================================================================
+//     Print
+// ================================================================================================
+
+// See genesis/utils/text/string.hpp
+// /**
+//  * @brief Print the bits in an int as a string.
+//  */
+// template <typename T>
+// std::string to_bit_string( T value, bool with_byte_spaces = true )
+// {
+//     static_assert(
+//         std::is_unsigned<T>::value, "Template parameter must be an unsigned integer type."
+//     );
+//     static_assert( CHAR_BIT == 8, "CHAR_BIT != 8" );
+//
+//     std::string res = "";
+//     T const one = 1;
+//     for( size_t i = 0; i < sizeof(T) * 8; ++i ) {
+//         res += (( value & (one << i)) ? "1" : "0");
+//         if( with_byte_spaces && (i+1) % 8 == 0 ) {
+//             res += " ";
+//         }
+//     }
+//     return res;
+// }
+
 // ================================================================================================
 //     Pop Count
 // ================================================================================================