Роутер работает нормально в process v2

This commit is contained in:
2026-04-07 14:09:51 +03:00
parent 0a25e42ea1
commit 8b7b72967e
1746 changed files with 216414 additions and 14037 deletions
+354 -5
View File
@@ -1,7 +1,7 @@
from __future__ import annotations
from app.modules.agent.intent_router_v2 import IntentRouterV2
from app.modules.agent.runtime.docs_qa_pipeline import DocsQAPipelineRunner
from app.core.agent.intent_router import IntentRouterV2
from app.core.agent.runtime.docs_qa_pipeline import DocsQAPipelineRunner, DocsTaskPlanner
from tests.docs_qa_eval.fixture_adapter import InMemoryDocsRetrievalAdapter
from tests.unit_tests.rag.intent_router_testkit import repo_context
@@ -135,6 +135,52 @@ def test_openapi_partial_contract_returns_partial_mode() -> None:
assert "/orders" in result.answer
def test_docs_pipeline_accepts_precomputed_task_plan_without_rerouting() -> None:
rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/api/health.md",
"title": "GET /health",
"content": "/health returns runtime and component statuses.",
"metadata": {"document_id": "api.health_endpoint", "type": "api_method", "endpoint": "/health"},
},
{
"layer": "D2_FACT_INDEX",
"path": "docs/api/health.md",
"title": "api.health_endpoint:response",
"content": "Returns health summary and component diagnostics.",
"metadata": {"subject_id": "api.health_endpoint", "type": "api_method"},
},
]
route_result = IntentRouterV2().route(
"Объясни API метод /health",
repo_context=repo_context(),
)
task_plan = DocsTaskPlanner().plan(
"Объясни API метод /health",
"docs-session",
route_result=route_result,
)
class FailingRouter:
def route(self, *_args, **_kwargs):
raise AssertionError("runner should use the precomputed task plan")
runner = DocsQAPipelineRunner(FailingRouter(), InMemoryDocsRetrievalAdapter(rows), repo_context=repo_context())
result = runner.run(
"Объясни API метод /health",
"docs-session",
mode="pre_llm_only",
task_plan=task_plan,
)
assert result.router_result.intent == "DOCUMENTATION_EXPLAIN"
assert result.router_result.query_plan.sub_intent == "API_METHOD_EXPLAIN"
assert result.diagnostics.selected_primary_documents == ["api.health_endpoint"]
assert result.diagnostics.gate_decision == "allow_exact"
def test_pre_llm_mode_returns_diagnostic_only_without_answer_generation() -> None:
rows = [
{
@@ -172,7 +218,7 @@ def test_pre_llm_mode_detects_path_anchor_candidates() -> None:
assert "/health" in result.diagnostics.query_anchor_candidates
assert "/health" in result.diagnostics.resolved_anchor_candidates
assert result.diagnostics.planned_layers == ["D2_FACT_INDEX", "D4_WORKFLOW_INDEX", "D1_DOCUMENT_CATALOG", "D0_DOC_CHUNKS"]
assert result.diagnostics.planned_layers == ["D1_DOCUMENT_CATALOG", "D2_FACT_INDEX", "D0_DOC_CHUNKS", "D4_WORKFLOW_INDEX"]
assert set(result.diagnostics.executed_layers) == {"D1_DOCUMENT_CATALOG", "D2_FACT_INDEX", "D4_WORKFLOW_INDEX", "D0_DOC_CHUNKS"}
@@ -318,8 +364,311 @@ def test_openapi_request_fragment_uses_fragment_aware_gate() -> None:
assert result.answer_mode in {"ready", "ready_partial"}
assert result.answer
assert "type: object" in result.answer
assert "message:" in result.answer
assert "chat_id:" in result.answer
def test_api_method_explain_prefers_api_method_primary_doc() -> None:
rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/domain/runtime-health.md",
"title": "Сущность runtime health",
"content": "Runtime health describes overall service health.",
"metadata": {"document_id": "domain.runtime_health", "type": "domain_entity"},
},
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/api/health.md",
"title": "GET /health",
"content": "/health returns runtime and component statuses.",
"metadata": {"document_id": "api.health_endpoint", "type": "api_method", "endpoint": "/health"},
},
{
"layer": "D5_RELATION_GRAPH",
"path": "docs/domain/runtime-health.md",
"title": "Runtime health links",
"content": "runtime health used by health endpoint",
"metadata": {"document_id": "domain.runtime_health", "target_doc_id": "api.health_endpoint"},
},
]
runner = DocsQAPipelineRunner(IntentRouterV2(), InMemoryDocsRetrievalAdapter(rows), repo_context=repo_context())
result = runner.run("Что делает метод health?", "docs-session", mode="pre_llm_only")
assert result.router_result.query_plan.sub_intent == "API_METHOD_EXPLAIN"
assert result.diagnostics.target_anchor in {"health", "/health"}
assert result.diagnostics.api_method_match_found is True
assert result.diagnostics.selected_primary_documents == ["api.health_endpoint"]
assert "api.health_endpoint" in result.diagnostics.primary_doc_candidates
assert result.diagnostics.evidence_gate_require_target_api_spec is True
assert result.diagnostics.evidence_gate_target_api_spec_found is True
assert result.answer_mode == "exact"
assert result.diagnostics.gate_decision == "allow_exact"
def test_api_method_explain_promotes_api_doc_via_links() -> None:
rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/domain/runtime-health.md",
"title": "Сущность runtime health",
"content": "Runtime health is the domain model for observability.",
"metadata": {
"document_id": "domain.runtime_health",
"type": "domain_entity",
"links": [{"target": "api.health_endpoint", "type": "used_by"}],
},
},
{
"layer": "D0_DOC_CHUNKS",
"path": "docs/api/health.md",
"title": "api.health_endpoint:Overview",
"content": "Endpoint /health returns overall runtime status and component diagnostics.",
"metadata": {"document_id": "api.health_endpoint", "type": "api_method", "endpoint": "/health"},
},
]
runner = DocsQAPipelineRunner(IntentRouterV2(), InMemoryDocsRetrievalAdapter(rows), repo_context=repo_context())
result = runner.run("Как работает health endpoint?", "docs-session", mode="pre_llm_only")
assert result.diagnostics.promoted_via_links == ["api.health_endpoint"]
assert result.diagnostics.selected_primary_documents == ["api.health_endpoint"]
assert result.diagnostics.api_method_match_found is True
def test_api_method_explain_rejects_cross_endpoint_primary_candidates() -> None:
rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/documentation/api/control-actions-endpoint.md",
"title": "HTTP API /actions/{action}",
"content": "Endpoint for controlling actions.",
"metadata": {
"document_id": "api.control_actions_endpoint",
"type": "api_method",
"endpoint": "/actions/{action}",
},
},
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/documentation/api/health-endpoint.md",
"title": "HTTP API /health",
"content": "Health endpoint returns runtime health and component diagnostics.",
"metadata": {
"document_id": "api.health_endpoint",
"type": "api_method",
"endpoint": "/health",
},
},
{
"layer": "D0_DOC_CHUNKS",
"path": "docs/documentation/api/actions-endpoint.md",
"title": "api.control_actions_endpoint:Scenario",
"content": "The /actions/{action} endpoint triggers runtime actions.",
"metadata": {
"document_id": "api.control_actions_endpoint",
"type": "api_method",
"endpoint": "/actions/{action}",
},
},
]
runner = DocsQAPipelineRunner(IntentRouterV2(), InMemoryDocsRetrievalAdapter(rows), repo_context=repo_context())
result = runner.run("Как работает метод health?", "docs-session", mode="pre_llm_only")
assert result.router_result.query_plan.sub_intent == "API_METHOD_EXPLAIN"
assert result.diagnostics.target_endpoint_identity["normalized_doc_id"] == "api.health_endpoint"
assert result.diagnostics.selected_primary_documents == ["api.health_endpoint"]
assert result.diagnostics.primary_api_documents_after_filter == ["api.health_endpoint"]
assert "api.control_actions_endpoint" in result.diagnostics.rejected_endpoint_candidates
assert result.diagnostics.cross_endpoint_leakage_detected is True
assert result.diagnostics.evidence_gate_target_api_spec_found is True
assert "api.control_actions_endpoint" not in result.diagnostics.selected_doc_ids
def test_api_method_explain_without_exact_target_returns_insufficiency() -> None:
rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/documentation/api/control-actions-endpoint.md",
"title": "HTTP API /actions/{action}",
"content": "Endpoint for controlling actions.",
"metadata": {
"document_id": "api.control_actions_endpoint",
"type": "api_method",
"endpoint": "/actions/{action}",
},
},
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/documentation/api/send-endpoint.md",
"title": "HTTP API /send",
"content": "Endpoint for sending messages.",
"metadata": {
"document_id": "api.send_message_endpoint",
"type": "api_method",
"endpoint": "/send",
},
},
]
runner = DocsQAPipelineRunner(IntentRouterV2(), InMemoryDocsRetrievalAdapter(rows), repo_context=repo_context())
result = runner.run("Что делает метод health?", "docs-session", mode="pre_llm_only")
assert result.router_result.query_plan.sub_intent == "API_METHOD_EXPLAIN"
assert result.diagnostics.target_endpoint_identity["normalized_doc_id"] == "api.health_endpoint"
assert result.diagnostics.selected_primary_documents == []
assert "api.control_actions_endpoint" in result.diagnostics.rejected_endpoint_candidates
assert "api.send_message_endpoint" in result.diagnostics.rejected_endpoint_candidates
assert result.diagnostics.target_api_spec_found_exact is False
assert result.diagnostics.evidence_gate_target_api_spec_found is False
assert result.diagnostics.gate_decision == "reject"
assert result.answer_mode == "insufficient"
assert "api.control_actions_endpoint" not in result.diagnostics.selected_doc_ids
assert "api.send_message_endpoint" not in result.diagnostics.selected_doc_ids
def test_api_method_explain_uses_indirect_mode_from_target_linked_docs() -> None:
rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/domain/runtime-health.md",
"title": "Runtime health",
"content": "Runtime health describes overall service state and component diagnostics.",
"metadata": {
"document_id": "domain.runtime_health",
"type": "domain_entity",
"links": [{"target": "api.health_endpoint", "type": "used_by"}],
},
},
]
runner = DocsQAPipelineRunner(IntentRouterV2(), InMemoryDocsRetrievalAdapter(rows), repo_context=repo_context())
result = runner.run("Как работает метод health?", "docs-session", mode="pre_llm_only")
assert result.answer_mode == "indirect"
assert result.diagnostics.gate_decision == "allow_indirect"
assert result.diagnostics.raw_retrieval_non_empty is True
assert result.diagnostics.target_primary_context_non_empty is False
assert result.diagnostics.indirect_target_context_non_empty is True
assert result.diagnostics.graph_promotion_attempted is True
assert result.diagnostics.graph_promotion_hits == ["api.health_endpoint"]
assert result.diagnostics.promoted_target_loaded is False
assert result.diagnostics.materialization_failure_reason == "materialized_rows_empty"
assert result.diagnostics.final_primary_document_ids == []
assert "domain.runtime_health" in result.diagnostics.final_secondary_document_ids
def test_api_method_explain_skips_llm_when_no_exact_or_indirect_context() -> None:
from tests.unit_tests.rag.test_docs_prompt_layer import FakeLlm
rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/documentation/api/send-endpoint.md",
"title": "HTTP API /send",
"content": "Endpoint for sending messages.",
"metadata": {
"document_id": "api.send_message_endpoint",
"type": "api_method",
"endpoint": "/send",
},
}
]
llm = FakeLlm("should not be called")
runner = DocsQAPipelineRunner(IntentRouterV2(), InMemoryDocsRetrievalAdapter(rows), repo_context=repo_context(), llm=llm)
result = runner.run("Что делает метод health?", "docs-session")
assert llm.calls == []
assert result.answer_mode == "insufficient"
assert result.diagnostics.llm_called is False
assert result.diagnostics.llm_call_reason == "no_exact_or_indirect_target_context"
assert result.diagnostics.gate_decision == "reject"
def test_api_method_explain_materializes_promoted_target_into_primary_context() -> None:
rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/domain/runtime-health.md",
"title": "Runtime health",
"content": "Runtime health describes service state and component diagnostics.",
"metadata": {
"document_id": "domain.runtime_health",
"type": "domain_entity",
"links": [{"target": "api.health_endpoint", "type": "used_by"}],
},
}
]
materialized_rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/api/health.md",
"title": "GET /health",
"content": "/health returns runtime and component statuses.",
"metadata": {"document_id": "api.health_endpoint", "type": "api_method", "endpoint": "/health"},
},
{
"layer": "D2_FACT_INDEX",
"path": "docs/api/health.md",
"title": "api.health_endpoint:response",
"content": "Returns health summary and component diagnostics.",
"metadata": {"subject_id": "api.health_endpoint", "type": "api_method"},
},
{
"layer": "D0_DOC_CHUNKS",
"path": "docs/api/health.md",
"title": "api.health_endpoint:Overview",
"content": "Endpoint /health returns overall runtime health.",
"metadata": {"document_id": "api.health_endpoint", "type": "api_method", "endpoint": "/health"},
},
]
runner = DocsQAPipelineRunner(
IntentRouterV2(),
InMemoryDocsRetrievalAdapter(rows, materialized_rows=materialized_rows),
repo_context=repo_context(),
)
result = runner.run("Как работает метод health?", "docs-session", mode="pre_llm_only")
assert result.answer_mode == "exact"
assert result.diagnostics.graph_promotion_hits == ["api.health_endpoint"]
assert result.diagnostics.graph_promotion_materialized == ["api.health_endpoint"]
assert result.diagnostics.promoted_target_loaded is True
assert result.diagnostics.promoted_target_chunks_loaded == 1
assert result.diagnostics.promoted_target_facts_loaded == 1
assert result.diagnostics.pinned_document_ids == ["api.health_endpoint"]
assert result.diagnostics.final_primary_document_ids == ["api.health_endpoint"]
assert "domain.runtime_health" in result.diagnostics.final_secondary_document_ids
assert result.diagnostics.materialized_target_primary_context_non_empty is True
assert result.diagnostics.gate_decision == "allow_exact"
def test_entity_question_does_not_prefer_api_method_primary_doc() -> None:
rows = [
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/domain/runtime-health.md",
"title": "Сущность runtime health",
"content": "Runtime health describes service state.",
"metadata": {"document_id": "domain.runtime_health", "type": "domain_entity"},
},
{
"layer": "D1_DOCUMENT_CATALOG",
"path": "docs/api/health.md",
"title": "GET /health",
"content": "/health returns runtime status.",
"metadata": {"document_id": "api.health_endpoint", "type": "api_method", "endpoint": "/health"},
},
]
runner = DocsQAPipelineRunner(IntentRouterV2(), InMemoryDocsRetrievalAdapter(rows), repo_context=repo_context())
result = runner.run("Что такое runtime health?", "docs-session", mode="pre_llm_only")
assert result.router_result.query_plan.sub_intent == "ENTITY_EXPLAIN"
assert result.diagnostics.selected_primary_documents == []
assert result.diagnostics.api_method_match_found is False
assert result.answer == ""
def test_openapi_method_with_only_path_is_rejected() -> None: