From 566130d1b350c61d940787b44f3fd3c678ded5a0 Mon Sep 17 00:00:00 2001
From: andrei <b1naryth1ef@gmail.com>
Date: Thu, 20 Jul 2017 16:26:44 -0700
Subject: [PATCH] [perf] greatly increase hashmap performance

UserDict was the wrong way to achieve this, modern versions of Python
can just subclass dict. This provides an immense performance boost by
allowing getitem/setitem calls to be routed directly to the underlying
storage within cpython land, instead of having to route through the
items MRO and eventually hit __dict__
---
 disco/util/hashmap.py | 16 ++++++----------
 tests/util/hashmap.py | 40 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 46 insertions(+), 10 deletions(-)
 create mode 100644 tests/util/hashmap.py

diff --git a/disco/util/hashmap.py b/disco/util/hashmap.py
index 50cf6f4..19a6b2d 100644
--- a/disco/util/hashmap.py
+++ b/disco/util/hashmap.py
@@ -1,21 +1,17 @@
 import six
 
-from six.moves import filter, map, UserDict
+from six.moves import filter, map
 from collections import defaultdict
 
 
-class HashMap(UserDict):
+class HashMap(dict):
+    __slots__ = ()
+
     def iter(self):
-        return iter(self.data)
+        return iter(self)
 
     def items(self):
-        return six.iteritems(self.data)
-
-    def keys(self):
-        return six.iterkeys(self.data)
-
-    def values(self):
-        return six.itervalues(self.data)
+        return six.iteritems(self)
 
     def find(self, predicate):
         if not callable(predicate):
diff --git a/tests/util/hashmap.py b/tests/util/hashmap.py
new file mode 100644
index 0000000..b57da6a
--- /dev/null
+++ b/tests/util/hashmap.py
@@ -0,0 +1,40 @@
+import pytest
+import random
+
+from disco.util.hashmap import HashMap
+
+
+@pytest.fixture
+def hashmap():
+    return HashMap({i: random.randint(1, 1000000) for i in range(100000)})
+
+
+def test_hashmap_insert_performance(benchmark):
+    def bench_hashmap_insert(hsh):
+        hsh[random.randint(1, 100000)] = True
+
+    benchmark(bench_hashmap_insert, HashMap())
+
+
+def test_hashmap_lookup_performance(benchmark, hashmap):
+    def bench_hashmap_lookup():
+        assert hashmap[random.randint(1, 10000)] > 0
+
+    benchmark(bench_hashmap_lookup)
+
+
+def test_hashmap_find(hashmap):
+    assert len(list(hashmap.find(lambda v: v > 0))) == len(hashmap)
+    assert hashmap.find_one(lambda v: v > 0) > 0
+
+
+def test_hashmap_filter(hashmap):
+    for item in list(hashmap.filter(lambda v: v % 2 == 0)):
+        assert item % 2 == 0
+
+
+def test_hashmap_builtins(hashmap):
+    for item in hashmap:
+        assert item in hashmap
+
+    assert hashmap.popitem()[1] > 1