multimap

Author: Ivan Butygin

sdc/_concurrent_hash.cpp

@@ -26,12 +26,11 @@
 
 #include <Python.h>
 #include <cstdint>
-#include <tbb/concurrent_hash_map.h>
-#include <tbb/concurrent_vector.h>
+#include <tbb/concurrent_unordered_map.h>
 
 
 template<typename Key, typename Val>
-using hashmap = tbb::concurrent_hash_map<Key,tbb::concurrent_vector<Val>>;
+using hashmap = tbb::concurrent_unordered_multimap<Key,Val>;
 
 template<typename Key, typename Val>
 using iter_range = std::pair<typename hashmap<Key, Val>::iterator, typename hashmap<Key, Val>::iterator>;
@@ -54,12 +53,7 @@ void delete_int_hashmap(void* obj)
 void addelem_int_hashmap(void* obj, int64_t key, size_t val)
 {
     auto& h = *static_cast<int_hashmap*>(obj);
-    int_hashmap::accessor ac;
-    h.insert(ac, key);
-    auto& vec = ac->second;
-    ac.release();
-    vec.push_back(val);
-    // h[key].push_back(val);
+    h.insert({key,val});
 }
 
 void* createiter_int_hashmap(void* obj)
@@ -86,16 +80,10 @@ int64_t iterkey_int_hashmap(void* it)
     return r.first->first;
 }
 
-size_t itersize_int_hashmap(void* it)
+size_t iterval_int_hashmap(void* it)
 {
     auto& r = *static_cast<int_hashmap_iters*>(it);
-    return r.first->second.size();
-}
-
-size_t iterelem_int_hashmap(void* it, size_t index)
-{
-    auto& r = *static_cast<int_hashmap_iters*>(it);
-    return r.first->second[index];
+    return r.first->second;
 }
 
 void deleteiter_int_hashmap(void* obj)
@@ -128,8 +116,7 @@ PyMODINIT_FUNC PyInit_hconcurrent_hash()
     REGISTER(enditer_int_hashmap)
     REGISTER(nextiter_int_hashmap)
     REGISTER(iterkey_int_hashmap)
-    REGISTER(itersize_int_hashmap)
-    REGISTER(iterelem_int_hashmap)
+    REGISTER(iterval_int_hashmap)
     REGISTER(deleteiter_int_hashmap)
 #undef REGISTER
     return m;

sdc/concurrent_hash.py

@@ -42,8 +42,7 @@
 ll.add_symbol('enditer_int_hashmap', hconcurrent_hash.enditer_int_hashmap)
 ll.add_symbol('nextiter_int_hashmap', hconcurrent_hash.nextiter_int_hashmap)
 ll.add_symbol('iterkey_int_hashmap', hconcurrent_hash.iterkey_int_hashmap)
-ll.add_symbol('itersize_int_hashmap', hconcurrent_hash.itersize_int_hashmap)
-ll.add_symbol('iterelem_int_hashmap', hconcurrent_hash.iterelem_int_hashmap)
+ll.add_symbol('iterval_int_hashmap', hconcurrent_hash.iterval_int_hashmap)
 ll.add_symbol('deleteiter_int_hashmap', hconcurrent_hash.deleteiter_int_hashmap)
 
 _create_int_hashmap = types.ExternalFunction("create_int_hashmap",
@@ -61,10 +60,8 @@
                                                types.void(types.voidptr))
 _iterkey_int_hashmap = types.ExternalFunction("iterkey_int_hashmap",
                                               types.int64(types.voidptr))
-_itersize_int_hashmap = types.ExternalFunction("itersize_int_hashmap",
-                                               types.intp(types.voidptr))
-_iterelem_int_hashmap = types.ExternalFunction("iterelem_int_hashmap",
-                                               types.intp(types.voidptr, types.intp))
+_iterval_int_hashmap = types.ExternalFunction("iterval_int_hashmap",
+                                              types.intp(types.voidptr))
 _deleteiter_int_hashmap = types.ExternalFunction("deleteiter_int_hashmap",
                                                  types.void(types.voidptr))
 
@@ -97,11 +94,7 @@ def iterkey_int_hashmap():
     pass
 
 
-def itersize_int_hashmap():
-    pass
-
-
-def iterelem_int_hashmap():
+def iterval_int_hashmap():
     pass
 
 
@@ -144,14 +137,9 @@ def iterkey_int_hashmap_overload(h):
     return lambda h: _iterkey_int_hashmap(h)
 
 
-@overload(itersize_int_hashmap)
-def itersize_int_hashmap_overload(h):
-    return lambda h: _itersize_int_hashmap(h)
-
-
-@overload(iterelem_int_hashmap)
-def iterelem_int_hashmap_overload(h, i):
-    return lambda h, i: _iterelem_int_hashmap(h, i)
+@overload(iterval_int_hashmap)
+def iterval_int_hashmap_overload(h):
+    return lambda h: _iterval_int_hashmap(h)
 
 
 @overload(deleteiter_int_hashmap)

sdc/tests/test_dataframe.py

@@ -1806,10 +1806,8 @@ def test_impl():
             it = sdc.concurrent_hash.createiter_int_hashmap(h)
             while 0 == sdc.concurrent_hash.enditer_int_hashmap(it):
                 key = sdc.concurrent_hash.iterkey_int_hashmap(it)
-                sz = sdc.concurrent_hash.itersize_int_hashmap(it)
-                for i in range(sz):
-                    val = sdc.concurrent_hash.iterelem_int_hashmap(it, i)
-                    print(key, val)
+                val = sdc.concurrent_hash.iterval_int_hashmap(it)
+                print(key, val)
 
                 sdc.concurrent_hash.nextiter_int_hashmap(it)