:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
from posthog.test.base import BaseTest, ClickhouseTestMixin
from unittest.mock import ANY, MagicMock, patch
from django.utils import timezone
from azure.ai.inference import EmbeddingsClient
from azure.ai.inference.models import EmbeddingItem, EmbeddingsResult, EmbeddingsUsage
from azure.core.credentials import AzureKeyCredential
from posthog.schema import MaxActionContext, MaxUIContext, TeamTaxonomyQuery
from posthog.hogql_queries.query_runner import ExecutionMode
from posthog.models.ai.utils import PgEmbeddingRow, bulk_create_pg_embeddings
from products.actions.backend.facade.testing import create_action
from ee.hogai.chat_agent.rag.nodes import InsightRagContextNode
from ee.hogai.utils.types import AssistantState
@patch(
"azure.ai.inference.EmbeddingsClient.embed",
return_value=EmbeddingsResult(
id="test",
model="test",
usage=EmbeddingsUsage(prompt_tokens=1, total_tokens=1),
data=[EmbeddingItem(embedding=[2, 4], index=0)],
),
)
@patch(
"ee.hogai.chat_agent.rag.nodes.get_azure_embeddings_client",
return_value=EmbeddingsClient(
endpoint="https://test.services.ai.azure.com/models", credential=AzureKeyCredential("test")
),
)
class TestInsightRagContextNode(ClickhouseTestMixin, BaseTest):
def setUp(self):
super().setUp()
self.action = create_action(
team=self.team,
name="Action",
description="Description",
summary="Summary",
last_summarized_at=timezone.now(),
embedding_last_synced_at=timezone.now(),
)
bulk_create_pg_embeddings(
[PgEmbeddingRow(domain="action", team_id=self.team.id, id=str(self.action.id), vector=[2, 4], text="test")]
)
@patch("ee.hogai.chat_agent.rag.nodes.TeamTaxonomyQueryRunner")
def test_prewarm_queries(self, mock_team_taxonomy_query_runner, cohere_mock, embed_mock):
# Arrange
team = MagicMock()
retriever = InsightRagContextNode(team=team, user=self.user)
mock_runner_instance = MagicMock()
mock_team_taxonomy_query_runner.return_value = mock_runner_instance
# Act
retriever._prewarm_queries()
# Assert
mock_team_taxonomy_query_runner.assert_called_once()
# Check TeamTaxonomyQueryRunner was instantiated with correct parameters
args, kwargs = mock_team_taxonomy_query_runner.call_args
assert isinstance(args[0], TeamTaxonomyQuery)
assert args[1] == team
# Check run was called with the correct execution mode
mock_runner_instance.run.assert_called_once_with(
ExecutionMode.RECENT_CACHE_CALCULATE_ASYNC_IF_STALE, user=self.user, analytics_props=ANY
)
def test_injects_action(self, cohere_mock, embed_mock):
retriever = InsightRagContextNode(team=self.team, user=self.user)
response = retriever.run(AssistantState(root_tool_insight_plan="Plan", messages=[]), {})
assert response is not None
assert response.rag_context is not None
self.assertIn("Action", response.rag_context)
self.assertIn("Description", response.rag_context)
self.assertIn(str(self.action.id), response.rag_context)
self.assertEqual(embed_mock.call_count, 1)
@patch.object(InsightRagContextNode, "context_manager", new_callable=lambda: MagicMock())
def test_injects_actions_from_context(self, mock_context_manager, cohere_mock, embed_mock):
# Create a second action that will come from UI context
context_action = create_action(
team=self.team,
name="Context Action",
description="From UI Context",
summary="Context Summary",
last_summarized_at=timezone.now(),
embedding_last_synced_at=timezone.now(),
)
# Mock UI context with actions
mock_ui_context = MaxUIContext(
actions=[MaxActionContext(id=context_action.id, name="Context Action", description="From UI Context")]
)
mock_context_manager.get_ui_context.return_value = mock_ui_context
retriever = InsightRagContextNode(team=self.team, user=self.user)
response = retriever.run(AssistantState(root_tool_insight_plan="Plan", messages=[]), {})
assert response is not None
assert response.rag_context is not None
# Should include both the vector-searched action and the context action
self.assertIn("Action", response.rag_context) # Original action from vector search
self.assertIn("Context Action", response.rag_context) # Action from UI context
self.assertIn("From UI Context", response.rag_context)
self.assertIn(str(context_action.id), response.rag_context)
self.assertEqual(embed_mock.call_count, 1)
@patch.object(InsightRagContextNode, "context_manager", new_callable=lambda: MagicMock())
def test_handles_actions_context_when_embedding_fails(self, mock_context_manager, cohere_mock, embed_mock):
# Make embedding fail
embed_mock.side_effect = ValueError("Embedding failed")
context_action = create_action(
team=self.team,
name="Context Only Action",
description="Only from context",
)
mock_ui_context = MaxUIContext(
actions=[
MaxActionContext(id=context_action.id, name="Context Only Action", description="Only from context")
]
)
mock_context_manager.get_ui_context.return_value = mock_ui_context
retriever = InsightRagContextNode(team=self.team, user=self.user)
response = retriever.run(AssistantState(root_tool_insight_plan="Plan", messages=[]), {})
assert response is not None
assert response.rag_context is not None
# Should still include actions from context even when embedding fails
self.assertIn("Context Only Action", response.rag_context)
self.assertIn("Only from context", response.rag_context)
self.assertIn(str(context_action.id), response.rag_context)