refactor: unroll varint encoding for more cases

Also:
- start using stream_data for property testing
- Varint.encode/1 no longer returns an iodata, but a binary

Benchmark shows a slight increase in performances and slight decrease in memory usage

lib/protox/varint.ex

@@ -4,23 +4,38 @@ defmodule Protox.Varint do
 
   import Bitwise
 
-  @spec encode(integer) :: iodata
-  def encode(v) when v < 128,
+  @spec encode(integer) :: binary()
+  def encode(v) when v < 1 <<< 7,
     do: <<v>>
 
-  def encode(v) when v < 16_384,
+  def encode(v) when v < 1 <<< 14,
     do: <<1::1, v::7, v >>> 7>>
 
-  def encode(v) when v < 2_097_152,
+  def encode(v) when v < 1 <<< 21,
     do: <<1::1, v::7, 1::1, v >>> 7::7, v >>> 14>>
 
-  def encode(v) when v < 268_435_456,
+  def encode(v) when v < 1 <<< 28,
     do: <<1::1, v::7, 1::1, v >>> 7::7, 1::1, v >>> 14::7, v >>> 21>>
 
-  def encode(v) when v < 34_359_738_368,
+  def encode(v) when v < 1 <<< 35,
     do: <<1::1, v::7, 1::1, v >>> 7::7, 1::1, v >>> 14::7, 1::1, v >>> 21::7, v >>> 28>>
 
-  def encode(v), do: [<<1::1, v::7>>, encode(v >>> 7)]
+  def encode(v) when v < 1 <<< 42,
+    do:
+      <<1::1, v::7, 1::1, v >>> 7::7, 1::1, v >>> 14::7, 1::1, v >>> 21::7, 1::1, v >>> 28::7,
+        v >>> 35>>
+
+  def encode(v) when v < 1 <<< 49,
+    do:
+      <<1::1, v::7, 1::1, v >>> 7::7, 1::1, v >>> 14::7, 1::1, v >>> 21::7, 1::1, v >>> 28::7,
+        1::1, v >>> 35::7, v >>> 42>>
+
+  def encode(v) when v < 1 <<< 56,
+    do:
+      <<1::1, v::7, 1::1, v >>> 7::7, 1::1, v >>> 14::7, 1::1, v >>> 21::7, 1::1, v >>> 28::7,
+        1::1, v >>> 35::7, 1::1, v >>> 42::7, v >>> 49>>
+
+  def encode(v), do: <<1::1, v::7, encode(v >>> 7)::binary>>
 
   @spec decode(binary) :: {non_neg_integer, binary}
   def decode(b), do: do_decode(0, 0, b)

mix.exs

@@ -39,7 +39,8 @@ defmodule Protox.Mixfile do
       {:dialyxir, "~> 1.0", only: [:test, :dev], runtime: false},
       {:excoveralls, "~> 0.13", only: [:test], runtime: false},
       {:ex_doc, "~> 0.22", only: [:dev], runtime: false},
-      {:propcheck, github: "alfert/propcheck", ref: "c564e89d", only: [:test, :dev]}
+      {:propcheck, github: "alfert/propcheck", ref: "c564e89d", only: [:test, :dev]},
+      {:stream_data, "~> 1.0", only: [:dev, :test], runtime: false}
     ]
     |> maybe_add_muzak_pro()
     |> maybe_download_protobuf()

mix.lock

@@ -22,4 +22,5 @@
   "proper": {:git, "https://github.com/proper-testing/proper.git", "a5ae5669f01143b0828fc21667d4f5e344aa760b", [ref: "a5ae5669f01143b0828fc21667d4f5e344aa760b"]},
   "protobuf": {:git, "https://github.com/protocolbuffers/protobuf.git", "b407e8416e3893036aee5af9a12bd9b6a0e2b2e6", [tag: "v29.3", submodules: true]},
   "statistex": {:hex, :statistex, "1.0.0", "f3dc93f3c0c6c92e5f291704cf62b99b553253d7969e9a5fa713e5481cd858a5", [:mix], [], "hexpm", "ff9d8bee7035028ab4742ff52fc80a2aa35cece833cf5319009b52f1b5a86c27"},
+  "stream_data": {:hex, :stream_data, "1.1.2", "05499eaec0443349ff877aaabc6e194e82bda6799b9ce6aaa1aadac15a9fdb4d", [:mix], [], "hexpm", "129558d2c77cbc1eb2f4747acbbea79e181a5da51108457000020a906813a1a9"},
 }

test/protox/varint_test.exs

@@ -1,41 +1,54 @@
 defmodule Protox.VarintTest do
-  use ExUnit.Case
-  use PropCheck
+  import Bitwise
+  use ExUnit.Case, async: true
+  use ExUnitProperties
+
+  property "Unrolled encoding produces the same result as the reference implementation" do
+    check all(int <- integer(0..(1 <<< 64))) do
+      assert int |> Protox.Varint.encode() |> IO.iodata_to_binary() ==
+               int |> encode_reference() |> IO.iodata_to_binary()
+    end
+  end
+
+  property "Symmetric" do
+    check all(int <- integer(0..(1 <<< 64))) do
+      assert int |> Protox.Varint.encode() |> IO.iodata_to_binary() |> Protox.Varint.decode() ==
+               {int, ""}
+    end
+  end
 
   test "Encode" do
     assert Protox.Varint.encode(0) == <<0>>
     assert Protox.Varint.encode(1) == <<1>>
 
-    assert Protox.Varint.encode(300) == <<172, 2>>
+    assert Protox.Varint.encode((1 <<< 14) - 1) == <<0xFF, 0x7F>>
+    assert Protox.Varint.encode(1 <<< 14) == <<0x80, 0x80, 0x1>>
 
-    assert Protox.Varint.encode(16_383) == <<0xFF, 0x7F>>
-    assert Protox.Varint.encode(16_384) == <<0x80, 0x80, 0x1>>
+    assert Protox.Varint.encode((1 <<< 21) - 1) == <<0xFF, 0xFF, 0x7F>>
+    assert Protox.Varint.encode(1 <<< 21) == <<0x80, 0x80, 0x80, 0x1>>
 
-    assert Protox.Varint.encode(2_097_151) == <<0xFF, 0xFF, 0x7F>>
-    assert Protox.Varint.encode(2_097_152) == <<0x80, 0x80, 0x80, 0x1>>
+    assert Protox.Varint.encode((1 <<< 28) - 1) == <<0xFF, 0xFF, 0xFF, 0x7F>>
+    assert Protox.Varint.encode(1 <<< 28) == <<0x80, 0x80, 0x80, 0x80, 0x1>>
 
-    assert Protox.Varint.encode(268_435_455) == <<0xFF, 0xFF, 0xFF, 0x7F>>
-    assert Protox.Varint.encode(268_435_456) == <<0x80, 0x80, 0x80, 0x80, 0x1>>
+    assert Protox.Varint.encode((1 <<< 35) - 1) == <<0xFF, 0xFF, 0xFF, 0xFF, 0x7F>>
+    assert Protox.Varint.encode(1 <<< 35) == <<0x80, 0x80, 0x80, 0x80, 0x80, 0x1>>
 
-    assert Protox.Varint.encode(34_359_738_367) == <<0xFF, 0xFF, 0xFF, 0xFF, 0x7F>>
-  end
+    assert Protox.Varint.encode((1 <<< 42) - 1) == <<0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x7F>>
+    assert Protox.Varint.encode(1 <<< 42) == <<0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x1>>
 
-  test "Decode" do
-    assert Protox.Varint.decode(<<172, 2>>) == {300, <<>>}
-    assert Protox.Varint.decode(<<172, 2, 0>>) == {300, <<0>>}
-    assert Protox.Varint.decode(<<0>>) == {0, <<>>}
-    assert Protox.Varint.decode(<<1>>) == {1, <<>>}
-    assert Protox.Varint.decode(<<185, 96>>) == {12_345, <<>>}
-    assert Protox.Varint.decode(<<185, 224, 0>>) == {12_345, <<>>}
-  end
+    assert Protox.Varint.encode((1 <<< 56) - 1) ==
+             <<0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x7F>>
 
-  @tag :properties
-  test "Symmetric" do
-    forall value <- integer() do
-      encoded = value |> Protox.Varint.encode() |> :binary.list_to_bin()
-      {decoded, <<>>} = Protox.Varint.decode(encoded)
+    assert Protox.Varint.encode(1 <<< 56) ==
+             <<0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x1>>
 
-      value == decoded
-    end
+    assert Protox.Varint.encode((1 <<< 63) - 1) ==
+             <<0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x7F>>
+
+    assert Protox.Varint.encode(1 <<< 63) ==
+             <<0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x1>>
   end
+
+  defp encode_reference(v) when v < 1 <<< 7, do: <<v>>
+  defp encode_reference(v), do: [<<1::1, v::7>>, encode_reference(v >>> 7)]
 end