1

ontology_platform/tests/unit/test_phase7_hybrid_extraction.py

@@ -1,6 +1,7 @@
 from crawler_platform.app.config.loader import ProjectConfig
+from crawler_platform.app.api.routes import extraction_log_summary
 from crawler_platform.app.core.extractor.base import ExtractedClaim, ExtractionBundle, ExtractionPageContext
-from crawler_platform.app.core.extractor.hybrid import HybridExtractor, mark_bundle, merge_bundles
+from crawler_platform.app.core.extractor.hybrid import HybridExtractor, fallback_bundle, mark_bundle, merge_bundles
 from crawler_platform.app.core.extractor.validation import validate_extraction_bundle
 
 
@@ -142,3 +143,61 @@ def test_hybrid_smart_routing_skips_llm_for_category_page() -> None:
     assert bundle.raw_output["llm_skipped"] is True
     assert bundle.raw_output["effective_extraction_mode"] == "rule_only"
     assert "CategoryPage" in bundle.raw_output["llm_skip_reason"]
+
+
+def test_rule_only_mode_records_rule_llm_summary_counts() -> None:
+    rule_bundle = ExtractionBundle(
+        claims=[
+            ExtractedClaim("A", "Thing", "color", object_value="red"),
+            ExtractedClaim("A", "Thing", "size", object_value="large"),
+        ]
+    )
+
+    class FixedRuleExtractor:
+        def extract(self, page_text, project_config):  # noqa: ANN001
+            return rule_bundle
+
+        def extract_from_context(self, context, project_config):  # noqa: ANN001
+            return rule_bundle
+
+    extractor = HybridExtractor("generic", mode="rule_only")
+    extractor.rule_extractor = FixedRuleExtractor()
+
+    bundle = extractor.extract("Alpha color red size large", project_config())
+
+    assert bundle.raw_output["rule_claim_count"] == 2
+    assert bundle.raw_output["llm_claim_count"] == 0
+    assert bundle.raw_output["comparison"]["rule_only"] == 2
+
+
+def test_fallback_records_rule_counts_and_comparison() -> None:
+    rule_bundle = ExtractionBundle(
+        claims=[
+            ExtractedClaim("A", "Thing", "color", object_value="red"),
+            ExtractedClaim("A", "Thing", "size", object_value="large"),
+        ]
+    )
+    mark_bundle(rule_bundle, source="rule", mode="hybrid")
+
+    bundle = fallback_bundle(rule_bundle, HybridExtractor("generic", mode="hybrid"), RuntimeError("boom"))
+
+    assert bundle.raw_output["rule_claim_count"] == 2
+    assert bundle.raw_output["llm_claim_count"] == 0
+    assert bundle.raw_output["comparison"]["rule_only"] == 2
+
+
+def test_extraction_log_summary_backfills_legacy_rule_only_counts() -> None:
+    summary = extraction_log_summary(
+        {
+            "validation": {"rejected_claim_count": 4},
+            "candidate_claims": [
+                {"metadata": {"agreement": "rule_only", "extraction_source": "rule"}},
+                {"metadata": {"agreement": "rule_only", "extraction_source": "rule"}},
+            ]
+        }
+    )
+
+    assert summary["rule_claim_count"] == 2
+    assert summary["llm_claim_count"] == 0
+    assert summary["comparison"]["rule_only"] == 2
+    assert summary["comparison"]["rejected_by_validation"] == 4