Add unit tests for GuidedReproduction

Author: TimeDelta

tests/test_guided_reproduction.py

@@ -0,0 +1,128 @@
+import os
+import sys
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(__file__)))
+import types
+
+import pytest
+
+from reproduction import GuidedReproduction
+
+
+class DummyGenome:
+    def __init__(self, key, fitness=0.0):
+        self.key = key
+        self.fitness = fitness
+        self.crossover_parents = None
+        self.mutated = False
+
+    def configure_crossover(self, p1, p2, config):
+        self.crossover_parents = (p1.key, p2.key)
+
+    def mutate(self, config):
+        self.mutated = True
+
+
+class DummySpecies:
+    def __init__(self, key, members):
+        self.key = key
+        self.members = {m.key: m for m in members}
+        self.adjusted_fitness = None
+
+
+class DummySpeciesSet:
+    def __init__(self, species):
+        self.species = {species.key: species}
+
+
+class DummyStagnation:
+    def update(self, species_set, generation):
+        return [(sid, s, False) for sid, s in species_set.species.items()]
+
+
+class DummyReporters:
+    def species_stagnant(self, *args, **kwargs):
+        pass
+
+    def info(self, *args, **kwargs):
+        pass
+
+
+class DummyConfig:
+    def __init__(self, genome_type):
+        self.genome_type = genome_type
+        self.genome_config = object()
+
+
+def make_reproduction(elitism=1, survival_threshold=0.5):
+    class RC:
+        def __init__(self):
+            self.elitism = elitism
+            self.survival_threshold = survival_threshold
+            self.min_species_size = 1
+
+    return GuidedReproduction(RC(), DummyReporters(), DummyStagnation())
+
+
+def test_guided_and_standard_offspring():
+    g1 = DummyGenome(101, fitness=1.0)
+    g2 = DummyGenome(102, fitness=0.5)
+    g3 = DummyGenome(103, fitness=0.2)
+    species = DummySpecies(1, [g1, g2, g3])
+    species_set = DummySpeciesSet(species)
+
+    repro = make_reproduction(elitism=1)
+    # force deterministic spawn amounts
+    repro.compute_spawn = lambda *a, **k: [3]
+
+    guided_called = {}
+
+    def guide_fn(name, features, starting, config, n_offspring):
+        guided_called["n"] = n_offspring
+        return [DummyGenome(50)]
+
+    repro.guide_fn = guide_fn
+
+    config = DummyConfig(DummyGenome)
+
+    class DummyTask:
+        def name(self):
+            return "dummy"
+
+        features = []
+
+    pop = repro.reproduce(config, species_set, 3, 0, DummyTask())
+
+    assert guided_called["n"] == 1
+    # expect 3 individuals: 1 elite, 1 guided, 1 crossover
+    assert len(pop) == 3
+    assert 101 in pop  # elite preserved
+    assert 50 in pop  # guided child
+    # crossover child gets key 1 from genome_indexer
+    assert any(isinstance(k, int) and k not in (101, 102, 103, 50) for k in pop)
+
+
+def test_only_elites_when_spawn_too_small():
+    g1 = DummyGenome(201, fitness=1.0)
+    g2 = DummyGenome(202, fitness=0.5)
+    species = DummySpecies(2, [g1, g2])
+    species_set = DummySpeciesSet(species)
+
+    repro = make_reproduction(elitism=1)
+    repro.compute_spawn = lambda *a, **k: [1]
+
+    repro.guide_fn = lambda *a, **k: pytest.fail("guide_fn should not be called")
+
+    config = DummyConfig(DummyGenome)
+
+    class DummyTask:
+        def name(self):
+            return "dummy"
+
+        features = []
+
+    pop = repro.reproduce(config, species_set, 2, 0, DummyTask())
+
+    # only elite should remain
+    assert len(pop) == 1
+    assert 201 in pop