Deduplication

tests/test_candidate_normalization.py

@@ -0,0 +1,89 @@
+from repo_registry.candidate_graph.generator import (
+    CandidateAbilityDraft,
+    CandidateCapabilityDraft,
+    CandidateFeatureDraft,
+)
+from repo_registry.candidate_graph.normalization import normalize_candidate_drafts
+from repo_registry.core.models import SourceReference
+
+
+def ref(fact_id, path):
+    return SourceReference(
+        fact_id=fact_id,
+        path=path,
+        kind="documentation",
+        name=path,
+    )
+
+
+def test_normalizer_merges_duplicate_abilities_and_nested_claims():
+    candidates = [
+        CandidateAbilityDraft(
+            name="LLM Provider Integration",
+            description="Connects to model providers.",
+            confidence=0.55,
+            source_refs=[ref(1, "README.md")],
+            capabilities=[
+                CandidateCapabilityDraft(
+                    name="Use OpenRouter Models",
+                    description="Calls OpenRouter.",
+                    inputs=["prompt"],
+                    outputs=["response"],
+                    confidence=0.6,
+                    source_refs=[ref(2, "providers.py")],
+                    features=[
+                        CandidateFeatureDraft(
+                            name="OpenRouter Model Support",
+                            type="integration",
+                            location="providers.py",
+                            confidence=0.6,
+                            source_refs=[ref(2, "providers.py")],
+                        )
+                    ],
+                )
+            ],
+        ),
+        CandidateAbilityDraft(
+            name="LLM Provider Integrations",
+            description="Connects prompts to OpenRouter and Claude providers.",
+            confidence=0.7,
+            source_refs=[ref(3, "providers.py")],
+            capabilities=[
+                CandidateCapabilityDraft(
+                    name="OpenRouter Model Support",
+                    description="Supports OpenRouter model calls.",
+                    inputs=["LLM request"],
+                    outputs=["model response"],
+                    confidence=0.75,
+                    source_refs=[ref(3, "providers.py")],
+                    features=[
+                        CandidateFeatureDraft(
+                            name="Use OpenRouter Models",
+                            type="backend integration",
+                            location="src/providers.py",
+                            confidence=0.75,
+                            source_refs=[ref(3, "providers.py")],
+                        )
+                    ],
+                )
+            ],
+        ),
+    ]
+
+    normalized = normalize_candidate_drafts(candidates)
+
+    assert len(normalized) == 1
+    ability = normalized[0]
+    assert ability.name == "LLM Provider Integrations"
+    assert ability.description == (
+        "Connects prompts to OpenRouter and Claude providers."
+    )
+    assert ability.confidence == 0.7
+    assert {ref.fact_id for ref in ability.source_refs} == {1, 3}
+    assert len(ability.capabilities) == 1
+    capability = ability.capabilities[0]
+    assert capability.confidence == 0.75
+    assert capability.inputs == ["prompt", "LLM request"]
+    assert capability.outputs == ["response", "model response"]
+    assert len(capability.features) == 1
+    assert capability.features[0].confidence == 0.75