Merge pull request #1 from findmyway/master

Use ReinforcementLearningBase as dep

.gitignore

@@ -1,3 +1,5 @@
 *.jl.cov
 *.jl.*.cov
 *.jl.mem
+
+Manifest.toml

.travis.yml

@@ -4,7 +4,6 @@ os:
   - linux
   - osx
 julia:
-  - 0.7
   - 1.0
   - nightly
 notifications:

Project.toml

@@ -0,0 +1,14 @@
+name = "ReinforcementLearningEnvironmentGym"
+uuid = "1053412c-b184-11e8-2eb2-ef51b44fed15"
+authors = ["Johanni Brea <[email protected]>", "Jun Tian <[email protected]>"]
+version = "0.1.0"
+
+[deps]
+PyCall = "438e738f-606a-5dbb-bf0a-cddfbfd45ab0"
+ReinforcementLearningBase = "e575027e-6cd6-5018-9292-cdc6200d2b44"
+
+[extras]
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[targets]
+test = ["Test"]

README.md

@@ -6,4 +6,4 @@
 
 [![codecov.io](http://codecov.io/github/JuliaReinforcementLearning/ReinforcementLearningEnvironmentGym.jl/coverage.svg?branch=master)](http://codecov.io/github/JuliaReinforcementLearning/ReinforcementLearningEnvironmentGym.jl?branch=master)
 
-Making the [OpenAI gym](https://github.com/openai/gym) environments available to the [Julia Reinforcement Learning](https://github.com/jbrea/ReinforcementLearning.jl) package.
+Making the [OpenAI gym](https://github.com/openai/gym) environments available to the [Julia Reinforcement Learning](https://github.com/JuliaReinforcementLearning/ReinforcementLearning.jl) package.

REQUIRE

@@ -1,4 +1,3 @@
-julia 0.7
+julia 1.0
 PyCall
-Reexport
-ReinforcementLearning
+ReinforcementLearningBase

src/ReinforcementLearningEnvironmentGym.jl

@@ -1,7 +1,7 @@
 module ReinforcementLearningEnvironmentGym
-using Reexport
-@reexport using ReinforcementLearning
-import ReinforcementLearning:interact!, reset!, getstate, plotenv
+export GymEnv, listallenvs, interact!, reset!, getstate, plotenv, actionspace, sample
+using ReinforcementLearningBase
+import ReinforcementLearningBase:interact!, reset!, getstate, plotenv, actionspace
 using PyCall
 const gym = PyNULL()
 
@@ -10,45 +10,52 @@ function __init__()
     pyimport("pybullet_envs")
 end
 
-function getspace(space)
-    if pyisinstance(space, gym[:spaces][:box][:Box])
-        ReinforcementLearning.Box(space[:low], space[:high])
-    elseif pyisinstance(space, gym[:spaces][:discrete][:Discrete])
-        1:space[:n]
-    else
-        error("Don't know how to convert $(pytypeof(space)).")
+function gymspace2jlspace(s::PyObject)
+    spacetype = s[:__class__][:__name__]
+    if     spacetype == "Box"           BoxSpace(s[:low], s[:high])
+    elseif spacetype == "Discrete"      DiscreteSpace(s[:n], 0)
+    elseif spacetype == "MultiBinary"   MultiBinarySpace(s[:n])
+    elseif spacetype == "MultiDiscrete" MultiDiscreteSpace(s[:nvec], 0)
+    elseif spacetype == "Tuple"         map(gymspace2jlspace, s[:spaces])
+    elseif spacetype == "Dict"          Dict(map((k, v) -> (k, gymspace2jlspace(v)), s[:spaces]))
+    else error("Don't know how to convert [$(spacetype)]")
     end
 end
-struct GymEnv{TObject, TObsSpace, TActionSpace}
-    pyobj::TObject
-    observation_space::TObsSpace
-    action_space::TActionSpace
+
+struct GymEnv <: AbstractEnv
+    pyobj::PyObject
+    observationspace::AbstractSpace
+    actionspace::AbstractSpace
     state::PyObject
 end
+
 function GymEnv(name::String)
     pyenv = gym[:make](name)
-    obsspace = getspace(pyenv[:observation_space])
-    actspace = getspace(pyenv[:action_space])
-    pyenv[:reset]()
+    obsspace = gymspace2jlspace(pyenv[:observation_space])
+    actspace = gymspace2jlspace(pyenv[:action_space])
     state = PyNULL()
-    pycall!(state, pyenv[:step], PyVector, pyenv[:action_space][:sample]())
-    pyenv[:reset]()
     GymEnv(pyenv, obsspace, actspace, state)
 end
 
-function interactgym!(action, env)
-    if env.state[3]
-        reset!(env)
-    end
+function interact!(env::GymEnv, action)
     pycall!(env.state, env.pyobj[:step], PyVector, action)
-    env.state[1], env.state[2], env.state[3]
+    (observation=env.state[1], reward=env.state[2], isdone=env.state[3])
+end
+
+"Not very useful, kept for compat"
+function getstate(env::GymEnv) 
+    if pyisinstance(env.state, PyCall.@pyglobalobj :PyTuple_Type)
+        (observation=env.state[1], isdone=env.state[3])
+    else
+        # env has just been reseted
+        (observation=env.state, isdone=false)
+    end
 end
-interact!(action, env::GymEnv) = interactgym!(action, env)
-interact!(action::Int64, env::GymEnv) = interactgym!(action - 1, env)
-reset!(env::GymEnv) = env.pyobj[:reset]()
-getstate(env::GymEnv) = (env.state[1], false) 
 
-plotenv(env::GymEnv, s, a, r, d) = env.pyobj[:render]()
+reset!(env::GymEnv) = (observation=pycall!(env.state, env.pyobj[:reset], PyArray),)
+plotenv(env::GymEnv) = env.pyobj[:render]()
+actionspace(env::GymEnv) = env.actionspace
+
 """
     listallenvs(pattern = r"")
 
@@ -64,6 +71,4 @@ function listallenvs(pattern = r"")
     end
 end
 
-export GymEnv, listallenvs
-
 end # module

test/runtests.jl

@@ -1,14 +1,9 @@
 using ReinforcementLearningEnvironmentGym
-@static if VERSION < v"0.7.0-DEV.2005"
-    using Base.Test
-else
-    using Test
-end
+using PyCall
+using Test
 
-import ReinforcementLearningEnvironmentGym: reset!, interact!, getstate
 for x in ["CartPole-v0"]
     env = GymEnv(x)
-    reset!(env)
-    @test typeof(interact!(1, env)) == Tuple{Array{Float64, 1}, Float64, Bool}
-    @test typeof(getstate(env)) == Tuple{Array{Float64, 1}, Bool}
+    @test typeof(reset!(env)) == NamedTuple{(:observation,), Tuple{PyArray{Float64, 1}}}
+    @test typeof(interact!(env, 1)) == NamedTuple{(:observation, :reward, :isdone), Tuple{Array{Float64, 1}, Float64, Bool}}
 end