feat: add knowledge import preflight

2026-09-10 20:22:53 +08:00
parent e9b3d272f0
commit 24f75636a0
3 changed files with 1488 additions and 0 deletions
@@ -0,0 +1,75 @@
+import json
+
+import pytest
+
+from tools.knowledge_import_preflight import build_import_manifest
+
+
+def vector_candidate() -> dict[str, object]:
+    return {
+        "qa_id": "RAG-PUB-001",
+        "title": "基金交易确认时间说明",
+        "question": "基金什么时候确认？",
+        "paraphrases": ["申购何时确认", "赎回多久确认"],
+        "answer": "交易确认时间以产品规则和实际交易日为准。",
+        "scope": "public",
+        "intent": "policy_explain",
+        "collection": "fin_policy_collection",
+        "execution_mode": "vector_search",
+        "retrieval_status": "approved_candidate",
+        "audience": ["visitor", "authenticated_user"],
+        "agent_data_access": "none",
+        "tags": ["交易规则", "确认"],
+        "phase": "phase_1",
+        "source_type": "qa_pair",
+        "source_file": "qa-v5.8.txt",
+        "source_url": None,
+        "source_version": "v5.8",
+        "review_status": "approved_candidate",
+        "status": "active",
+        "effective_date": None,
+        "expire_date": None,
+    }
+
+
+def rule_only_record() -> dict[str, object]:
+    return {
+        "qa_id": "RAG-SEC-001",
+        "scope": "security_notice",
+        "collection": None,
+        "execution_mode": "fixed_route",
+        "retrieval_status": "rule_only",
+    }
+
+
+def test_manifest_contains_only_eligible_public_records_pending_admin_review() -> None:
+    manifest = build_import_manifest(
+        [vector_candidate(), rule_only_record()], source_name="qa-v5.8.jsonl"
+    )
+
+    assert manifest["summary"] == {
+        "total_records": 2,
+        "eligible_records": 1,
+        "excluded_rule_records": 1,
+        "publication_state": "pending_review",
+    }
+    entry = manifest["records"][0]
+    assert entry["qa_id"] == "RAG-PUB-001"
+    assert entry["knowledge_type"] == "policy_explain"
+    assert entry["milvus_collection"] == "fin_policy_collection"
+    assert entry["review_status"] == "pending_review"
+    assert entry["status"] == "active"
+    assert entry["retrieval_text"] == (
+        "标准问题：基金什么时候确认？\n"
+        "相似问法：申购何时确认；赎回多久确认\n"
+        "标签：交易规则、确认"
+    )
+    assert json.loads(entry["content_text"])["answer"] == "交易确认时间以产品规则和实际交易日为准。"
+
+
+def test_invalid_public_record_is_rejected_instead_of_silently_entering_manifest() -> None:
+    invalid = vector_candidate()
+    invalid["agent_data_access"] = "account"
+
+    with pytest.raises(ValueError, match="RAG-PUB-001"):
+        build_import_manifest([invalid], source_name="qa-v5.8.jsonl")