Make python classes available in Julia

test/Project.toml

@@ -4,6 +4,7 @@ authors = ["Alinson S. Xavier <git@axavier.org>"]
 version = "0.1.0"
 
 [deps]
+Glob = "c27321d9-0574-5035-807b-f59d2c89b15c"
 HDF5 = "f67ccb44-e63f-5c2f-98bd-6dc0ccc4ba2f"
 HiGHS = "87dc4568-4c63-4d18-b0c0-bb2238e4078b"
 JuMP = "4076af6c-e467-56ae-b986-b466b2749572"

test/src/MIPLearnT.jl

@@ -4,21 +4,26 @@ using Test
 using Logging
 using JuliaFormatter
 using HiGHS
+using Glob
 
 BASEDIR = dirname(@__FILE__)
 FIXTURES = "$BASEDIR/../fixtures"
 
+include("fixtures.jl")
+
 include("Cuts/BlackBox/test_cplex.jl")
 include("problems/test_setcover.jl")
-include("test_h5.jl")
+include("test_io.jl")
 include("solvers/test_jump.jl")
+include("test_usage.jl")
 
 function runtests()
     @testset "MIPLearn" begin
         test_cuts_blackbox_cplex()
-        test_h5()
+        test_io()
         test_problems_setcover()
         test_solvers_jump()
+        test_usage()
     end
 end
 

test/src/fixtures.jl

@@ -0,0 +1,14 @@
+function fixture_setcover_data()
+    return SetCoverData(
+        costs = [5, 10, 12, 6, 8],
+        incidence_matrix = [
+            1 0 0 1 0
+            1 1 0 0 0
+            0 0 1 1 1
+        ],
+    )
+end
+
+function fixture_setcover_model()
+    return build_setcover_model(fixture_setcover_data())
+end

test/src/solvers/test_jump.jl

@@ -1,18 +1,6 @@
 using JuMP
 import MIPLearn: from_str_array, to_str_array
 
-function build_model()
-    data = SetCoverData(
-        costs = [5, 10, 12, 6, 8],
-        incidence_matrix = [
-            1 0 0 1 0
-            1 1 0 0 0
-            0 0 1 1 1
-        ],
-    )
-    return build_setcover_model(data)
-end
-
 function test_solvers_jump()
     test_solvers_jump_extract()
     test_solvers_jump_add_constrs()
@@ -51,7 +39,7 @@ function test_solvers_jump_extract()
         @test all(actual .≈ expected)
     end
 
-    model = build_model()
+    model = fixture_setcover_model()
     model.extract_after_load(h5)
     test_sparse(
         "static_constr_lhs",
@@ -106,7 +94,7 @@ end
 
 function test_solvers_jump_add_constrs()
     h5 = H5File(tempname(), "w")
-    model = build_model()
+    model = fixture_setcover_model()
     model.extract_after_load(h5)
     model.add_constrs(
         to_str_array(["x[2]", "x[3]"]),
@@ -124,12 +112,9 @@ end
 
 function test_solvers_jump_fix_vars()
     h5 = H5File(tempname(), "w")
-    model = build_model()
+    model = fixture_setcover_model()
     model.extract_after_load(h5)
-    model.fix_variables(
-        to_str_array(["x[2]", "x[3]"]),
-        [0, 0],
-    )
+    model.fix_variables(to_str_array(["x[2]", "x[3]"]), [0, 0])
     model.optimize()
     model.extract_after_mip(h5)
     @test all(h5.get_array("mip_var_values") .≈ [1, 0, 0, 0, 1])
@@ -138,7 +123,7 @@ end
 function test_solvers_jump_warm_starts()
     # TODO: Check presence of warm start on log file
     h5 = H5File(tempname(), "w")
-    model = build_model()
+    model = fixture_setcover_model()
     model.extract_after_load(h5)
     model.set_warm_starts(
         to_str_array(["x[0]", "x[1]", "x[2]", "x[3]", "x[4]"]),
@@ -149,8 +134,8 @@ end
 
 function test_solvers_jump_write()
     mps_filename = "$(tempname()).mps"
-    model = build_model()
+    model = fixture_setcover_model()
     model.write(mps_filename)
     @test isfile(mps_filename)
     rm(mps_filename)
-end
+end

test/src/test_io.jl

@@ -1,5 +1,18 @@
 using MIPLearn
 
+function test_io()
+    test_pkl_gz()
+    test_h5()
+end
+
+function test_pkl_gz()
+    original = Dict("K1" => 1, "K2" => [0, 1, 2], "K3" => "Hello")
+    dirname = tempdir()
+    MIPLearn.write_pkl_gz([original], dirname)
+    recovered = MIPLearn.read_pkl_gz("$dirname/00000.pkl.gz")
+    @test recovered == original
+end
+
 function test_h5()
     h5 = H5File(tempname(), "w")
     _test_roundtrip_scalar(h5, "A")

test/src/test_usage.jl

@@ -0,0 +1,40 @@
+
+function test_usage()
+    LogisticRegression = pyimport("sklearn.linear_model").LogisticRegression
+
+    @debug "Generating data files..."
+    dirname = tempdir()
+    data = [fixture_setcover_data()]
+    data_filenames = write_pkl_gz(data, dirname)
+    h5_filenames = ["$(f).h5" for f in data_filenames]
+
+    @debug "Setting up LearningSolver..."
+    solver = LearningSolver(
+        components = [
+            IndependentVarsPrimalComponent(
+                base_clf = SingleClassFix(
+                    MinProbabilityClassifier(
+                        base_clf = LogisticRegression(),
+                        thresholds = [0.95, 0.95],
+                    ),
+                ),
+                extractor = AlvLouWeh2017Extractor(),
+                action = SetWarmStart(),
+            ),
+        ],
+    )
+
+    @debug "Collecting training data..."
+    bc = BasicCollector()
+    bc.collect(data_filenames, build_setcover_model)
+
+    @debug "Training models..."
+    solver.fit(data_filenames)
+
+    @debug "Solving model..."
+    solver.optimize(data_filenames[1], build_setcover_model)
+
+    @debug "Checking solution..."
+    h5 = H5File(h5_filenames[1])
+    @test h5.get_scalar("mip_obj_value") == 11.0
+end