Remove dense_hash_map/set cruft

This implements a general wrapper for dense_hash_map and dense_hash_set,
that falls back to unordered_map and unordered_set, if sparsehash is not
enabled.

src/graph/Makefile.am

@@ -56,6 +56,7 @@ libgraph_tool_core_la_include_HEADERS = \
     graph_python_interface.hh \
     graph_selectors.hh \
     graph_util.hh \
+    hash_map_wrap.hh \
     histogram.hh \
     mpl_nested_loop.hh \
     numpy_bind.hh \

src/graph/clustering/graph_clustering.hh

@@ -20,13 +20,9 @@
 
 #include "config.h"
 
-#include <unordered_set>
+#include "hash_map_wrap.hh"
 #include <boost/mpl/if.hpp>
 
-#ifdef HAVE_SPARSEHASH
-#include SPARSEHASH_INCLUDE(dense_hash_set)
-#endif
-
 #ifndef __clang__
 #include <ext/numeric>
 using __gnu_cxx::power;
@@ -42,12 +38,6 @@ namespace graph_tool
 {
 using namespace boost;
 
-#ifdef HAVE_SPARSEHASH
-using google::dense_hash_set;
-#else
-using std::unordered_set;
-#endif
-
 // calculates the number of triangles to which v belongs
 template <class Graph>
 pair<int,int>
@@ -55,18 +45,11 @@ get_triangles(typename graph_traits<Graph>::vertex_descriptor v, const Graph &g)
 {
     typedef typename graph_traits<Graph>::vertex_descriptor vertex_t;
 
-#ifdef HAVE_SPARSEHASH
-    typedef dense_hash_set<vertex_t, std::hash<vertex_t>> set_t;
-#else
-    typedef unordered_set<vertex_t> set_t;
-#endif
+    typedef gt_hash_set<vertex_t> set_t;
 
     set_t neighbour_set;
 
-#ifdef HAVE_SPARSEHASH
-     neighbour_set.set_empty_key(numeric_limits<vertex_t>::max());
-     neighbour_set.resize(out_degree(v, g));
-#endif
+    neighbour_set.resize(out_degree(v, g));
 
     size_t triangles = 0;
 

src/graph/community/graph_blockmodel.cc

@@ -603,11 +603,7 @@ struct get_deg_entropy_term_overlap
     void operator()(Graph& g, Vprop b, overlap_stats_t& overlap_stats, size_t N,
                     double& S) const
     {
-#ifdef HAVE_SPARSEHASH
-        typedef dense_hash_map<int, int, std::hash<int>> map_t;
-#else
-        typedef unordered_map<int, int> map_t;
-#endif
+        typedef gt_hash_map<int, int> map_t;
 
         map_t in_hist, out_hist;
 

src/graph/community/graph_blockmodel.hh

@@ -18,21 +18,16 @@
 #ifndef GRAPH_BLOCKMODEL_HH
 #define GRAPH_BLOCKMODEL_HH
 
+#include "config.h"
+
 #include <cmath>
 #include <iostream>
 #include <queue>
 #include <boost/math/special_functions/zeta.hpp>
 #include <boost/functional/hash.hpp>
-
-#include "config.h"
-#include <unordered_set>
-#include <unordered_map>
 #include <tuple>
 
-#ifdef HAVE_SPARSEHASH
-#include SPARSEHASH_INCLUDE(dense_hash_set)
-#include SPARSEHASH_INCLUDE(dense_hash_map)
-#endif
+#include "hash_map_wrap.hh"
 
 #include "../generation/sampler.hh"
 #include "../generation/dynamic_sampler.hh"
@@ -47,11 +42,6 @@ double spence(double);
 namespace graph_tool
 {
 
-#ifdef HAVE_SPARSEHASH
-using google::dense_hash_set;
-using google::dense_hash_map;
-#endif
-
 using namespace boost;
 
 template <class Key>
@@ -382,11 +372,7 @@ class partition_stats_t
 {
 public:
 
-#ifdef HAVE_SPARSEHASH
-    typedef dense_hash_map<pair<size_t,size_t>, int, std::hash<pair<size_t,size_t>>> map_t;
-#else
-    typedef unordered_map<pair<size_t,size_t>, int> map_t;
-#endif
+    typedef gt_hash_map<pair<size_t,size_t>, int> map_t;
 
     partition_stats_t() : _enabled(false) {}
 
@@ -396,17 +382,6 @@ public:
         : _enabled(true), _N(N), _E(0), _B(B), _hist(B), _total(B), _ep(B),
           _em(B), _edges_dl(edges_dl)
     {
-
-#ifdef HAVE_SPARSEHASH
-        for (size_t r = 0; r < B; ++r)
-        {
-            _hist[r].set_empty_key(make_pair(numeric_limits<size_t>::max(),
-                                             numeric_limits<size_t>::max()));
-            _hist[r].set_deleted_key(make_pair(numeric_limits<size_t>::max() - 1,
-                                               numeric_limits<size_t>::max() - 1));
-        }
-#endif
-
         for (auto v : vertices_range(g))
         {
             auto r = b[v];
@@ -719,12 +694,8 @@ struct get_ehash_t
     {
         typedef typename graph_traits<Graph>::vertex_descriptor vertex_t;
         typedef typename graph_traits<Graph>::edge_descriptor edge_t;
-#ifdef HAVE_SPARSEHASH
-        typedef dense_hash_map<vertex_t, edge_t, std::hash<vertex_t>> map_t;
-#else
-        typedef unordered_map<vertex_t, edge_t> map_t;
-#endif
-        typedef vector<map_t> type;
+        typedef gt_hash_map<vertex_t, edge_t> ehash_t;
+        typedef std::vector<ehash_t> type;
     };
 };
 
@@ -782,25 +753,11 @@ struct create_ehash
     void operator()(Graph& g, boost::any& oemap) const
     {
         typedef typename get_ehash_t::apply<Graph>::type emat_t;
-        typedef typename graph_traits<Graph>::vertex_descriptor vertex_t;
-
         emat_t emat(num_vertices(g));
 
-#ifdef HAVE_SPARSEHASH
-        for (auto v : vertices_range(g))
-        {
-            emat[v].set_empty_key(numeric_limits<vertex_t>::max());
-            emat[v].set_deleted_key(numeric_limits<vertex_t>::max() - 1);
-        }
-#endif
-
         for (auto e : edges_range(g))
             put_me(source(e, g), target(e, g), e, emat, g);
 
-#ifdef HAVE_SPARSEHASH
-        for (auto v : vertices_range(g))
-            emat[v].resize(0);
-#endif
         oemap = emat;
     }
 };

src/graph/community/graph_blockmodel_covariates.hh

@@ -23,25 +23,15 @@
 #include <queue>
 #include <boost/math/special_functions/zeta.hpp>
 #include <boost/functional/hash.hpp>
+#include <tuple>
 
 #include "config.h"
-#include <unordered_set>
-#include <unordered_map>
-#include <tuple>
 
-#ifdef HAVE_SPARSEHASH
-#include SPARSEHASH_INCLUDE(dense_hash_set)
-#include SPARSEHASH_INCLUDE(dense_hash_map)
-#endif
+#include "hash_map_wrap.hh"
 
 namespace graph_tool
 {
 
-#ifdef HAVE_SPARSEHASH
-using google::dense_hash_set;
-using google::dense_hash_map;
-#endif
-
 using namespace boost;
 
 // this will label each edge covariate in a consecutive range [0, C-1]
@@ -104,17 +94,7 @@ struct split_graph
                 size_t u_r;
 
                 if (block_map.size() <= l + 1)
-                {
-                    size_t n = block_map.size();
                     block_map.resize(l + 2);
-#ifdef HAVE_SPARSEHASH
-                    for (size_t i = n; i < block_map.size(); ++i)
-                    {
-                        block_map[i].set_empty_key(numeric_limits<size_t>::max());
-                        block_map[i].set_deleted_key(numeric_limits<size_t>::max() - 1);
-                    }
-#endif
-                }
                 auto& bmap = block_map[l + 1];
                 auto riter = bmap.find(r);
                 if (riter == bmap.end())
@@ -150,12 +130,7 @@ struct split_graph
     }
 };
 
-#ifdef HAVE_SPARSEHASH
-    typedef vector<dense_hash_map<size_t, size_t, std::hash<size_t>>> bmap_t;
-#else
-    typedef vector<unordered_map<size_t, size_t>> bmap_t;
-#endif
-
+typedef vector<gt_hash_map<size_t, size_t>> bmap_t;
 
 template <class BlockState>
 size_t get_block_map(BlockState& state, typename bmap_t::value_type& bmap,

src/graph/community/graph_blockmodel_overlap.hh

@@ -25,17 +25,8 @@
 
 #include "graph_blockmodel.hh"
 
-#ifdef HAVE_SPARSEHASH
-#include SPARSEHASH_INCLUDE(dense_hash_set)
-#include SPARSEHASH_INCLUDE(dense_hash_map)
-#endif
-
 namespace graph_tool
 {
-#ifdef HAVE_SPARSEHASH
-using google::dense_hash_set;
-using google::dense_hash_map;
-#endif
 
 using namespace boost;
 
@@ -359,12 +350,7 @@ public:
     bool is_enabled() const { return _enabled; }
 
 
-#ifdef HAVE_SPARSEHASH
-    typedef dense_hash_map<pair<size_t, size_t>, int,
-                           std::hash<pair<size_t, size_t>>> phist_t;
-#else
-    typedef unordered_map<pair<size_t, size_t>, int> phist_t;
-#endif
+    typedef gt_hash_map<pair<size_t, size_t>, int> phist_t;
 
     const vector<phist_t>& get_parallel_bundles() { return _parallel_bundles; }
     const vector<int>& get_mi() { return _mi; }
@@ -374,11 +360,7 @@ private:
     vvmap_t _half_edges;     // half-edges to each node
     vimap_t _node_index;     // node to each half edges
 
-#ifdef HAVE_SPARSEHASH
-    typedef dense_hash_map<size_t, deg_t, std::hash<size_t>> node_map_t;
-#else
-    typedef unordered_map<size_t, deg_t> node_map_t;
-#endif
+    typedef gt_hash_map<size_t, deg_t> node_map_t;
 
     vector<node_map_t> _block_nodes; // nodes (and degrees) in each block
 
@@ -405,47 +387,24 @@ struct overlap_partition_stats_t
 
     typedef vector<int> bv_t;
 
-#ifdef HAVE_SPARSEHASH
-    typedef dense_hash_map<bv_t, size_t, std::hash<bv_t>> bhist_t;
-    typedef dense_hash_map<cdeg_t, size_t, std::hash<cdeg_t>> cdeg_hist_t;
+    typedef gt_hash_map<bv_t, size_t> bhist_t;
+    typedef gt_hash_map<cdeg_t, size_t, std::hash<cdeg_t>> cdeg_hist_t;
 
-    typedef dense_hash_map<bv_t, cdeg_hist_t, std::hash<bv_t>> deg_hist_t;
+    typedef gt_hash_map<bv_t, cdeg_hist_t> deg_hist_t;
 
-    typedef dense_hash_map<bv_t, vector<size_t>, std::hash<bv_t>> ebhist_t;
-#else
-    typedef unordered_map<bv_t, size_t> bhist_t;
-    typedef unordered_map<cdeg_t, size_t, std::hash<cdeg_t>> cdeg_hist_t;
+    typedef gt_hash_map<bv_t, vector<size_t>> ebhist_t;
 
-    typedef unordered_map<bv_t, cdeg_hist_t> deg_hist_t;
-
-    typedef unordered_map<bv_t, vector<size_t>> ebhist_t;
-#endif
-
-    typedef unordered_map<int, int> dmap_t;
+    typedef gt_hash_map<int, int> dmap_t;
 
     overlap_partition_stats_t() : _enabled(false) {}
 
     template <class Graph, class Vprop, class Eprop>
     overlap_partition_stats_t(Graph& g, Vprop b, overlap_stats_t& overlap_stats,
                               Eprop eweight, size_t N, size_t B, bool edges_dl)
-        : _enabled(true), _N(N), _B(B), _D(0),
-          _dhist(B + 1), _r_count(B), _bhist(N), _emhist(B), _ephist(B),
-          _embhist(N), _epbhist(N), _deg_hist(N), _bvs(N), _nbvs(N), _degs(N),
-          _ndegs(N), _deg_delta(B), _edges_dl(edges_dl)
+        : _enabled(true), _N(N), _B(B), _D(0), _dhist(B + 1), _r_count(B),
+          _bvs(N), _nbvs(N), _degs(N), _ndegs(N), _deg_delta(B),
+          _edges_dl(edges_dl)
     {
-#ifdef HAVE_SPARSEHASH
-        bv_t empty = {numeric_limits<int>::max()};
-        bv_t deleted = {numeric_limits<int>::max() - 1};
-        _bhist.set_empty_key(empty);
-        _bhist.set_deleted_key(deleted);
-        _deg_hist.set_empty_key(empty);
-        _deg_hist.set_deleted_key(deleted);
-        _embhist.set_empty_key(empty);
-        _embhist.set_deleted_key(deleted);
-        _epbhist.set_empty_key(empty);
-        _epbhist.set_deleted_key(deleted);
-#endif
-
         dmap_t in_hist, out_hist;
         for (size_t v = 0; v < N; ++v)
         {
@@ -473,18 +432,6 @@ struct overlap_partition_stats_t
             _ndegs[v].reserve(cdeg.size() * 2);
 
             auto & cdh = _deg_hist[bv];
-#ifdef HAVE_SPARSEHASH
-            if (cdh.empty())
-            {
-                cdeg_t empty = {deg_t(numeric_limits<int>::max(),
-                                      numeric_limits<int>::max())};
-                cdeg_t deleted = {deg_t(numeric_limits<int>::max() - 1,
-                                        numeric_limits<int>::max() - 1)};
-                cdh.set_empty_key(empty);
-                cdh.set_deleted_key(deleted);
-            }
-
-#endif
             cdh[cdeg]++;
 
             size_t d = bv.size();
@@ -1017,15 +964,8 @@ struct overlap_partition_stats_t
             S_b += lbinom_fast(_r_count[s] + _ephist[s] - 1, _ephist[s]);
         }
 
-#ifdef HAVE_SPARSEHASH
-        dense_hash_map<size_t, pair<int, int>, std::hash<size_t>> deg_delta;
-        dense_hash_map<size_t, int, std::hash<size_t>> r_count_delta;
-        deg_delta.set_empty_key(numeric_limits<size_t>::max());
-        r_count_delta.set_empty_key(numeric_limits<size_t>::max());
-#else
-        unordered_map<size_t, pair<int, int>> deg_delta;
-        unordered_map<size_t, int> r_count_delta;
-#endif
+        gt_hash_map<size_t, pair<int, int>> deg_delta;
+        gt_hash_map<size_t, int> r_count_delta;
 
         if (bv != n_bv)
         {
@@ -1171,17 +1111,6 @@ struct overlap_partition_stats_t
         _ephist[r] -= kout;
 
         auto& hist = _deg_hist[n_bv];
-#ifdef HAVE_SPARSEHASH
-        if (hist.empty())
-        {
-            cdeg_t empty = {deg_t(numeric_limits<int>::max(),
-                                  numeric_limits<int>::max())};
-            cdeg_t deleted = {deg_t(numeric_limits<int>::max() - 1,
-                                    numeric_limits<int>::max() - 1)};
-            hist.set_empty_key(empty);
-            hist.set_deleted_key(deleted);
-        }
-#endif
         hist[n_deg] += 1;
         auto& n_bmh = _embhist[n_bv];
         auto& n_bph = _epbhist[n_bv];

src/graph/community/graph_community_network.hh

@@ -18,11 +18,7 @@
 #ifndef GRAPH_COMMUNITY_NETWORK_HH
 #define GRAPH_COMMUNITY_NETWORK_HH
 
-#include <unordered_map>
-
-#ifdef HAVE_SPARSEHASH
-#include SPARSEHASH_INCLUDE(dense_hash_map)
-#endif
+#include "hash_map_wrap.hh"
 
 #include <iostream>
 #include <iomanip>
@@ -50,17 +46,12 @@ struct get_community_network_vertices
         typedef typename boost::property_traits<CommunityMap>::value_type
             s_type;
 
-#ifdef HAVE_SPARSEHASH
-        google::dense_hash_map<s_type, vertex_t, std::hash<s_type> > comms;
-        comms.set_empty_key(numeric_limits<s_type>::max());
-#else
-        std::unordered_map<s_type, vertex_t, std::hash<s_type> > comms;
-#endif
+        gt_hash_map<s_type, vertex_t, std::hash<s_type> > comms;
+
         // create vertices
-        typename graph_traits<Graph>::vertex_iterator vi, vi_end;
-        for (tie(vi, vi_end) = vertices(g); vi != vi_end; ++vi)
+        for (auto vi : vertices_range(g))
         {
-            s_type s = get(s_map, *vi);
+            s_type s = get(s_map, vi);
             typeof(comms.begin()) iter = comms.find(s);
             cvertex_t v;
             if (iter == comms.end())
@@ -75,7 +66,7 @@ struct get_community_network_vertices
             {
                 v = iter->second;
             }
-            put(vertex_count, v, get(vertex_count, v) + get(vweight, *vi));
+            put(vertex_count, v, get(vertex_count, v) + get(vweight, vi));
         }
     }
 
@@ -116,37 +107,22 @@ struct get_community_network_edges
         typedef typename boost::property_traits<CommunityMap>::value_type
             s_type;
 
-#ifdef HAVE_SPARSEHASH
-        typedef google::dense_hash_map<s_type, vertex_t, std::hash<s_type> > comms_t;
-        comms_t comms(num_vertices(cg));
-        comms.set_empty_key(numeric_limits<s_type>::max());
-        typedef google::dense_hash_map<cvertex_t, cedge_t, std::hash<cvertex_t>> ecomms_t;
-#else
-        typedef std::unordered_map<s_type, vertex_t, std::hash<s_type> > comms_t;
-        comms_t comms(num_vertices(cg));
-        typedef std::unordered_map<cvertex_t, cedge_t> ecomms_t;
-#endif
+        typedef gt_hash_map<s_type, vertex_t, std::hash<s_type> > comms_t;
+        comms_t comms;
+        typedef gt_hash_map<cvertex_t, cedge_t> ecomms_t;
 
         auto index_map = get(vertex_index_t(), cg);
         unchecked_vector_property_map<ecomms_t, decltype(index_map)>
             comm_edges(index_map, num_vertices(cg));
 
-        typename graph_traits<CommunityGraph>::vertex_iterator v, v_end;
-        for (tie(v, v_end) = vertices(cg); v != v_end; ++v)
-        {
-            comms[cs_map[*v]] = *v;
-#ifdef HAVE_SPARSEHASH
-            comm_edges[*v].set_empty_key(numeric_limits<cvertex_t>::max());
-#endif
-        }
-
+        for (auto v : vertices_range(cg))
+            comms[cs_map[v]] = v;
 
         // create edges
-        typename graph_traits<Graph>::edge_iterator e, e_end;
-        for (tie(e, e_end) = edges(g); e != e_end; ++e)
+        for (auto e : edges_range(g))
         {
-            cvertex_t cs = comms[get(s_map, source(*e, g))];
-            cvertex_t ct = comms[get(s_map, target(*e, g))];
+            cvertex_t cs = comms[get(s_map, source(e, g))];
+            cvertex_t ct = comms[get(s_map, target(e, g))];
             if (ct == cs && !self_loops)
                 continue;
             cedge_t ce;
@@ -157,7 +133,7 @@ struct get_community_network_edges
             }
             else
             {
-                typeof(comm_edges[cs].begin()) iter = comm_edges[cs].find(ct);
+                auto iter = comm_edges[cs].find(ct);
                 if (iter != comm_edges[cs].end())
                 {
                     ce = iter->second;
@@ -184,7 +160,7 @@ struct get_community_network_edges
                     }
                 }
             }
-            put(edge_count, ce, get(edge_count, ce) + get(eweight, *e));
+            put(edge_count, ce, get(edge_count, ce) + get(eweight, e));
         }
     }
 };
@@ -241,9 +217,8 @@ struct get_weighted_vertex_property
     void operator()(const Graph& g, VertexWeightMap vweight, Vprop vprop,
                     Vprop temp) const
     {
-        typename graph_traits<Graph>::vertex_iterator vi, vi_end;
-        for (tie(vi, vi_end) = vertices(g); vi != vi_end; ++vi)
-            temp[*vi] = vprop[*vi] * get(vweight, *vi);
+        for (auto vi : vertices_range(g))
+            temp[vi] = vprop[vi] * get(vweight, vi);
     }
 };
 
@@ -259,21 +234,15 @@ struct get_vertex_community_property_sum
         typedef typename boost::property_traits<CommunityMap>::value_type
             s_type;
 
-#ifdef HAVE_SPARSEHASH
-        google::dense_hash_map<s_type, vertex_t, std::hash<s_type> > comms(num_vertices(cg));
-        comms.set_empty_key(numeric_limits<s_type>::max());
-#else
-        std::unordered_map<s_type, vertex_t, std::hash<s_type> > comms(num_vertices(cg));
-#endif
-        typename graph_traits<CommunityGraph>::vertex_iterator v, v_end;
-        for (tie(v, v_end) = vertices(cg); v != v_end; ++v)
-            comms[cs_map[*v]] = *v;
-
-        typename graph_traits<Graph>::vertex_iterator vi, vi_end;
-        for (tie(vi, vi_end) = vertices(g); vi != vi_end; ++vi)
+        gt_hash_map<s_type, vertex_t> comms;
+
+        for (auto v : vertices_range(cg))
+            comms[cs_map[v]] = v;
+
+        for (auto v : vertices_range(g))
         {
-            s_type s = get(s_map, *vi);
-            cvprop[comms[s]] += vprop[*vi];
+            s_type s = get(s_map, v);
+            cvprop[comms[s]] += vprop[v];
         }
     }
 };
@@ -284,9 +253,8 @@ struct get_weighted_edge_property
     void operator()(const Graph& g, EdgeWeightMap eweight, Eprop eprop,
                     Eprop temp) const
     {
-        typename graph_traits<Graph>::edge_iterator ei, ei_end;
-        for (tie(ei, ei_end) = edges(g); ei != ei_end; ++ei)
-            temp[*ei] = eprop[*ei] * get(eweight, *ei);
+        for (auto e : edges_range(g))
+            temp[e] = eprop[e] * get(eweight, e);
     }
 };
 
@@ -306,43 +274,27 @@ struct get_edge_community_property_sum
         typedef typename boost::property_traits<CommunityMap>::value_type
             s_type;
 
-#ifdef HAVE_SPARSEHASH
-        google::dense_hash_map<s_type, vertex_t, std::hash<s_type> > comms(num_vertices(cg));
-        comms.set_empty_key(numeric_limits<s_type>::max());
-#else
-        std::unordered_map<s_type, vertex_t, std::hash<s_type> > comms(num_vertices(cg));
-#endif
-        typename graph_traits<CommunityGraph>::vertex_iterator v, v_end;
-        for (tie(v, v_end) = vertices(cg); v != v_end; ++v)
-            comms[cs_map[*v]] = *v;
-
-#ifdef HAVE_SPARSEHASH
-        google::dense_hash_map<pair<size_t, size_t>, cedge_t,
-                               std::hash<pair<size_t, size_t> > >
-             comm_edges(num_vertices(cg));
-        comm_edges.set_empty_key(make_pair(numeric_limits<size_t>::max(),
-                                           numeric_limits<size_t>::max()));
-#else
-        std::unordered_map<pair<size_t, size_t>, cedge_t>
-            comm_edges(num_vertices(cg));
-#endif
-
-        typename graph_traits<CommunityGraph>::edge_iterator e, e_end;
-        for (tie(e, e_end) = edges(cg); e != e_end; ++e)
+        gt_hash_map<s_type, vertex_t> comms;
+
+        for (auto v : vertices_range(cg))
+            comms[cs_map[v]] = v;
+
+        gt_hash_map<pair<size_t, size_t>, cedge_t> comm_edges;
+
+        for (auto e : edges_range(cg))
         {
-            cvertex_t cs = comms[get(cs_map, source(*e, cg))];
-            cvertex_t ct = comms[get(cs_map, target(*e, cg))];
-            comm_edges[make_pair(cs, ct)] = *e;
+            cvertex_t cs = comms[get(cs_map, source(e, cg))];
+            cvertex_t ct = comms[get(cs_map, target(e, cg))];
+            comm_edges[make_pair(cs, ct)] = e;
         }
 
-        typename graph_traits<Graph>::edge_iterator ei, ei_end;
-        for (tie(ei, ei_end) = edges(g); ei != ei_end; ++ei)
+        for (auto e : edges_range(g))