Fix rerun assessment and candidate extraction

2026-05-16 00:57:44 +02:00
parent bee770fad7
commit 3e906c1dd4
7 changed files with 227 additions and 12 deletions
--- a/tests/test_candidate_graph.py
+++ b/tests/test_candidate_graph.py
@@ -135,6 +135,85 @@ def test_candidate_generator_extracts_intended_capability_blocks_from_intent_chu
    assert [ref.path for ref in intent_capability.source_refs] == ["INTENT.md"]


+def test_candidate_generator_preserves_unicode_and_normalizes_analysis_names():
+    repository = Repository(
+        id=1,
+        name="VergabeTeilnahme",
+        url="/tmp/vergabe-teilnahme",
+        description=None,
+        branch="main",
+        status="analyzed",
+    )
+    facts = [
+        fact(
+            1,
+            "intent",
+            "INTENT",
+            "INTENT.md",
+            metadata={"source_role": "intent_summary"},
+        )
+    ]
+    chunks = [
+        chunk(
+            1,
+            "intent",
+            "INTENT.md",
+            "# INTENT\n\n"
+            "Vollständiger Implementierungsplan in 12 Ralph-Loop-Workplans.\n\n"
+            "## Intended Capabilities\n\n"
+            "- Analysis of impact risk and dependency chains.\n",
+        )
+    ]
+
+    graph = CandidateGraphGenerator().generate(repository, facts, chunks)
+
+    assert graph[0].name == "Vollständiger Implementierungsplan In 12 Ralph-Loop-Workplans"
+    assert graph[0].capabilities[0].name == "Analyze Impact Risk And Dependency Chains"
+
+
+def test_candidate_generator_extracts_primary_outcome_subsections_from_intent():
+    repository = Repository(
+        id=1,
+        name="HelixForge",
+        url="/tmp/helix-forge",
+        description=None,
+        branch="main",
+        status="analyzed",
+    )
+    facts = [
+        fact(
+            1,
+            "intent",
+            "INTENT",
+            "INTENT.md",
+            metadata={"source_role": "intent_summary"},
+        )
+    ]
+    chunks = [
+        chunk(
+            1,
+            "intent",
+            "INTENT.md",
+            "# INTENT\n\n"
+            "HelixForge turns intent into structure.\n\n"
+            "## 4\\. Primary outcomes\n\n"
+            "### 4.1 Capability discovery\n\n"
+            "Clarify scope and ownership.\n\n"
+            "### 4.2 Capability validation\n\n"
+            "Validate architecture descriptions structurally and semantically.\n\n"
+            "## Architectural foundation\n\n"
+            "This section should not become a capability.\n",
+        )
+    ]
+
+    graph = CandidateGraphGenerator().generate(repository, facts, chunks)
+
+    capability_names = {capability.name for capability in graph[0].capabilities}
+    assert "Support Capability Discovery" in capability_names
+    assert "Validate Capabilities" in capability_names
+    assert "Architectural Foundation" not in capability_names
+
+
 def test_candidate_generator_prefers_intent_over_derived_scope_for_ability_name():
    repository = Repository(
        id=1,
@@ -238,7 +317,7 @@ def test_candidate_generator_extracts_current_capabilities_from_scope_blocks():
            "scope",
            "SCOPE.md",
            "# SCOPE\n\n## One-liner\n"
-            "S5 Workloads and Experience layer of the Railiance OAS Stack.\n\n"
+            "S5 Workloads and Experience layer of the Railiance OAS Stack -- owns applications.\n\n"
            "## Provided Capabilities\n\n"
            "```capability\n"
            "type: infrastructure\n"
@@ -255,6 +334,7 @@ def test_candidate_generator_extracts_current_capabilities_from_scope_blocks():

    ability = graph[0]
    assert ability.name == "S5 Workloads And Experience Layer Of The Railiance OAS Stack"
+    assert ability.name == "S5 Workloads And Experience Layer Of The Railiance OAS Stack"
    capability = ability.capabilities[0]
    assert capability.name == "Application workload deployment"
    assert capability.primary_class == "infrastructure"
--- a/tests/test_cli.py
+++ b/tests/test_cli.py
@@ -226,6 +226,19 @@ def test_assess_dataset_cli_reports_sparse_hierarchy_issues(tmp_path):
    )
    repository = service.register_repository(name="Scope Only", url=str(source))
    service.analyze_repository(repository.id, use_llm_assistance=False)
+    (source / "SCOPE.md").write_text(
+        "# SCOPE\n\n"
+        "## One-liner\n"
+        "Scope-only current behavior.\n\n"
+        "## Provided Capabilities\n\n"
+        "```capability\n"
+        "name: Review Latest Scope Facts\n"
+        "type: scope-review\n"
+        "description: Review the latest scope facts instead of stale runs.\n"
+        "```\n",
+        encoding="utf-8",
+    )
+    latest_summary = service.analyze_repository(repository.id, use_llm_assistance=False)
    output_path = tmp_path / "dataset.json"

    exit_code = main(
@@ -247,6 +260,7 @@ def test_assess_dataset_cli_reports_sparse_hierarchy_issues(tmp_path):
    assert exit_code == 0
    assert report["schema_version"] == "repo-scoping-dataset-assessment/v1"
    assert repo_report["name"] == "Scope Only"
+    assert repo_report["latest_analysis_run_id"] == latest_summary.analysis_run.id
    assert repo_report["documents"]["SCOPE.md"] is True
    assert repo_report["candidate_counts"]["capabilities"] >= 1
    assert repo_report["dependency_graph"]["node_count"] > 0
--- a/tests/test_registry_service.py
+++ b/tests/test_registry_service.py
@@ -541,6 +541,50 @@ def test_dependency_graph_renders_candidate_fallback_when_approved_hierarchy_mis
    assert any(edge["dependencyType"] == "draft-supports" for edge in edges)


+def test_dependency_graph_candidate_fallback_uses_latest_completed_run(tmp_path):
+    service = make_service(tmp_path)
+    source = tmp_path / "latest-scope-candidate"
+    source.mkdir()
+    (source / "SCOPE.md").write_text(
+        "# SCOPE\n\n## One-liner\nOld scope summary.\n",
+        encoding="utf-8",
+    )
+    repository = service.register_repository(name="Latest Scope Candidate", url=str(source))
+    service.analyze_repository(
+        repository.id,
+        source_path=str(source),
+        use_llm_assistance=False,
+    )
+    (source / "SCOPE.md").write_text(
+        "# SCOPE\n\n"
+        "## One-liner\n"
+        "Latest scope summary.\n\n"
+        "## Provided Capabilities\n\n"
+        "```capability\n"
+        "type: review\n"
+        "title: Latest Scope Capability\n"
+        "description: The second run should drive graph fallback.\n"
+        "```\n",
+        encoding="utf-8",
+    )
+    latest = service.analyze_repository(
+        repository.id,
+        source_path=str(source),
+        use_llm_assistance=False,
+    )
+
+    payload = service.dependency_graph_elements(repository.id, use_latest_profile=False)
+
+    labels = {
+        element["data"].get("label")
+        for element in payload["elements"]
+        if "source" not in element["data"]
+    }
+    assert latest.analysis_run.id == service.list_analysis_runs(repository.id)[0].id
+    assert "Latest Scope Capability" in labels
+    assert "Old Scope Summary" not in labels
+
+
 def test_manual_registry_updates_and_deletes_approved_entries(tmp_path):
    service = make_service(tmp_path)
    repository = service.register_repository(