235 lines
8 KiB
Python
235 lines
8 KiB
Python
|
|
"""Tests for agents/runner.py — agent execution with mocked claude CLI."""
|
||
|
|
|
||
|
|
import json
|
||
|
|
import pytest
|
||
|
|
from unittest.mock import patch, MagicMock
|
||
|
|
from core.db import init_db
|
||
|
|
from core import models
|
||
|
|
from agents.runner import run_agent, run_pipeline, _try_parse_json
|
||
|
|
|
||
|
|
|
||
|
|
@pytest.fixture
|
||
|
|
def conn():
|
||
|
|
c = init_db(":memory:")
|
||
|
|
models.create_project(c, "vdol", "ВДОЛЬ", "~/projects/vdolipoperek",
|
||
|
|
tech_stack=["vue3"])
|
||
|
|
models.create_task(c, "VDOL-001", "vdol", "Fix bug",
|
||
|
|
brief={"route_type": "debug"})
|
||
|
|
yield c
|
||
|
|
c.close()
|
||
|
|
|
||
|
|
|
||
|
|
def _mock_claude_success(output_data):
|
||
|
|
"""Create a mock subprocess result with successful claude output."""
|
||
|
|
mock = MagicMock()
|
||
|
|
mock.stdout = json.dumps(output_data) if isinstance(output_data, dict) else output_data
|
||
|
|
mock.stderr = ""
|
||
|
|
mock.returncode = 0
|
||
|
|
return mock
|
||
|
|
|
||
|
|
|
||
|
|
def _mock_claude_failure(error_msg):
|
||
|
|
mock = MagicMock()
|
||
|
|
mock.stdout = ""
|
||
|
|
mock.stderr = error_msg
|
||
|
|
mock.returncode = 1
|
||
|
|
return mock
|
||
|
|
|
||
|
|
|
||
|
|
# ---------------------------------------------------------------------------
|
||
|
|
# run_agent
|
||
|
|
# ---------------------------------------------------------------------------
|
||
|
|
|
||
|
|
class TestRunAgent:
|
||
|
|
@patch("agents.runner.subprocess.run")
|
||
|
|
def test_successful_agent_run(self, mock_run, conn):
|
||
|
|
mock_run.return_value = _mock_claude_success({
|
||
|
|
"result": "Found race condition in useSearch.ts",
|
||
|
|
"usage": {"total_tokens": 5000},
|
||
|
|
"cost_usd": 0.015,
|
||
|
|
})
|
||
|
|
|
||
|
|
result = run_agent(conn, "debugger", "VDOL-001", "vdol")
|
||
|
|
|
||
|
|
assert result["success"] is True
|
||
|
|
assert result["role"] == "debugger"
|
||
|
|
assert result["model"] == "sonnet"
|
||
|
|
assert result["duration_seconds"] >= 0
|
||
|
|
|
||
|
|
# Verify claude was called with right args
|
||
|
|
call_args = mock_run.call_args
|
||
|
|
cmd = call_args[0][0]
|
||
|
|
assert "claude" in cmd[0]
|
||
|
|
assert "-p" in cmd
|
||
|
|
assert "--output-format" in cmd
|
||
|
|
assert "json" in cmd
|
||
|
|
|
||
|
|
@patch("agents.runner.subprocess.run")
|
||
|
|
def test_failed_agent_run(self, mock_run, conn):
|
||
|
|
mock_run.return_value = _mock_claude_failure("API error")
|
||
|
|
|
||
|
|
result = run_agent(conn, "debugger", "VDOL-001", "vdol")
|
||
|
|
|
||
|
|
assert result["success"] is False
|
||
|
|
|
||
|
|
# Should be logged in agent_logs
|
||
|
|
logs = conn.execute("SELECT * FROM agent_logs WHERE task_id='VDOL-001'").fetchall()
|
||
|
|
assert len(logs) == 1
|
||
|
|
assert logs[0]["success"] == 0
|
||
|
|
|
||
|
|
def test_dry_run_returns_prompt(self, conn):
|
||
|
|
result = run_agent(conn, "debugger", "VDOL-001", "vdol", dry_run=True)
|
||
|
|
|
||
|
|
assert result["dry_run"] is True
|
||
|
|
assert result["prompt"] is not None
|
||
|
|
assert "VDOL-001" in result["prompt"]
|
||
|
|
assert result["output"] is None
|
||
|
|
|
||
|
|
@patch("agents.runner.subprocess.run")
|
||
|
|
def test_agent_logs_to_db(self, mock_run, conn):
|
||
|
|
mock_run.return_value = _mock_claude_success({"result": "ok"})
|
||
|
|
|
||
|
|
run_agent(conn, "tester", "VDOL-001", "vdol")
|
||
|
|
|
||
|
|
logs = conn.execute("SELECT * FROM agent_logs WHERE agent_role='tester'").fetchall()
|
||
|
|
assert len(logs) == 1
|
||
|
|
assert logs[0]["project_id"] == "vdol"
|
||
|
|
|
||
|
|
@patch("agents.runner.subprocess.run")
|
||
|
|
def test_previous_output_passed(self, mock_run, conn):
|
||
|
|
mock_run.return_value = _mock_claude_success({"result": "tests pass"})
|
||
|
|
|
||
|
|
run_agent(conn, "tester", "VDOL-001", "vdol",
|
||
|
|
previous_output="Found bug in line 42")
|
||
|
|
|
||
|
|
call_args = mock_run.call_args
|
||
|
|
prompt = call_args[0][0][2] # -p argument
|
||
|
|
assert "line 42" in prompt
|
||
|
|
|
||
|
|
|
||
|
|
# ---------------------------------------------------------------------------
|
||
|
|
# run_pipeline
|
||
|
|
# ---------------------------------------------------------------------------
|
||
|
|
|
||
|
|
class TestRunPipeline:
|
||
|
|
@patch("agents.runner.subprocess.run")
|
||
|
|
def test_successful_pipeline(self, mock_run, conn):
|
||
|
|
mock_run.return_value = _mock_claude_success({"result": "done"})
|
||
|
|
|
||
|
|
steps = [
|
||
|
|
{"role": "debugger", "brief": "find bug"},
|
||
|
|
{"role": "tester", "depends_on": "debugger", "brief": "verify"},
|
||
|
|
]
|
||
|
|
result = run_pipeline(conn, "VDOL-001", steps)
|
||
|
|
|
||
|
|
assert result["success"] is True
|
||
|
|
assert result["steps_completed"] == 2
|
||
|
|
assert len(result["results"]) == 2
|
||
|
|
|
||
|
|
# Pipeline created in DB
|
||
|
|
pipe = conn.execute("SELECT * FROM pipelines WHERE task_id='VDOL-001'").fetchone()
|
||
|
|
assert pipe is not None
|
||
|
|
assert pipe["status"] == "completed"
|
||
|
|
|
||
|
|
# Task updated to review
|
||
|
|
task = models.get_task(conn, "VDOL-001")
|
||
|
|
assert task["status"] == "review"
|
||
|
|
|
||
|
|
@patch("agents.runner.subprocess.run")
|
||
|
|
def test_pipeline_fails_on_step(self, mock_run, conn):
|
||
|
|
# First step succeeds, second fails
|
||
|
|
mock_run.side_effect = [
|
||
|
|
_mock_claude_success({"result": "found bug"}),
|
||
|
|
_mock_claude_failure("compilation error"),
|
||
|
|
]
|
||
|
|
|
||
|
|
steps = [
|
||
|
|
{"role": "debugger", "brief": "find"},
|
||
|
|
{"role": "frontend_dev", "brief": "fix"},
|
||
|
|
{"role": "tester", "brief": "test"},
|
||
|
|
]
|
||
|
|
result = run_pipeline(conn, "VDOL-001", steps)
|
||
|
|
|
||
|
|
assert result["success"] is False
|
||
|
|
assert result["steps_completed"] == 1 # Only debugger completed
|
||
|
|
assert "frontend_dev" in result["error"]
|
||
|
|
|
||
|
|
# Pipeline marked as failed
|
||
|
|
pipe = conn.execute("SELECT * FROM pipelines WHERE task_id='VDOL-001'").fetchone()
|
||
|
|
assert pipe["status"] == "failed"
|
||
|
|
|
||
|
|
# Task marked as blocked
|
||
|
|
task = models.get_task(conn, "VDOL-001")
|
||
|
|
assert task["status"] == "blocked"
|
||
|
|
|
||
|
|
def test_pipeline_dry_run(self, conn):
|
||
|
|
steps = [
|
||
|
|
{"role": "debugger", "brief": "find"},
|
||
|
|
{"role": "tester", "brief": "verify"},
|
||
|
|
]
|
||
|
|
result = run_pipeline(conn, "VDOL-001", steps, dry_run=True)
|
||
|
|
|
||
|
|
assert result["dry_run"] is True
|
||
|
|
assert result["success"] is True
|
||
|
|
assert result["steps_completed"] == 2
|
||
|
|
|
||
|
|
# No pipeline created in DB
|
||
|
|
pipes = conn.execute("SELECT * FROM pipelines").fetchall()
|
||
|
|
assert len(pipes) == 0
|
||
|
|
|
||
|
|
@patch("agents.runner.subprocess.run")
|
||
|
|
def test_pipeline_chains_output(self, mock_run, conn):
|
||
|
|
"""Output from step N is passed as previous_output to step N+1."""
|
||
|
|
call_count = [0]
|
||
|
|
|
||
|
|
def side_effect(*args, **kwargs):
|
||
|
|
call_count[0] += 1
|
||
|
|
if call_count[0] == 1:
|
||
|
|
return _mock_claude_success({"result": "bug is in line 42"})
|
||
|
|
return _mock_claude_success({"result": "test written"})
|
||
|
|
|
||
|
|
mock_run.side_effect = side_effect
|
||
|
|
|
||
|
|
steps = [
|
||
|
|
{"role": "debugger", "brief": "find"},
|
||
|
|
{"role": "tester", "brief": "write test"},
|
||
|
|
]
|
||
|
|
run_pipeline(conn, "VDOL-001", steps)
|
||
|
|
|
||
|
|
# Second call should include first step's output in prompt
|
||
|
|
second_call = mock_run.call_args_list[1]
|
||
|
|
prompt = second_call[0][0][2] # -p argument
|
||
|
|
assert "line 42" in prompt or "bug" in prompt
|
||
|
|
|
||
|
|
def test_pipeline_task_not_found(self, conn):
|
||
|
|
result = run_pipeline(conn, "NONEXISTENT", [{"role": "debugger"}])
|
||
|
|
assert result["success"] is False
|
||
|
|
assert "not found" in result["error"]
|
||
|
|
|
||
|
|
|
||
|
|
# ---------------------------------------------------------------------------
|
||
|
|
# JSON parsing
|
||
|
|
# ---------------------------------------------------------------------------
|
||
|
|
|
||
|
|
class TestTryParseJson:
|
||
|
|
def test_direct_json(self):
|
||
|
|
assert _try_parse_json('{"a": 1}') == {"a": 1}
|
||
|
|
|
||
|
|
def test_json_in_code_fence(self):
|
||
|
|
text = 'Some text\n```json\n{"a": 1}\n```\nMore text'
|
||
|
|
assert _try_parse_json(text) == {"a": 1}
|
||
|
|
|
||
|
|
def test_json_embedded_in_text(self):
|
||
|
|
text = 'Here is the result: {"status": "ok", "count": 42} and more'
|
||
|
|
result = _try_parse_json(text)
|
||
|
|
assert result == {"status": "ok", "count": 42}
|
||
|
|
|
||
|
|
def test_empty_string(self):
|
||
|
|
assert _try_parse_json("") is None
|
||
|
|
|
||
|
|
def test_no_json(self):
|
||
|
|
assert _try_parse_json("just plain text") is None
|
||
|
|
|
||
|
|
def test_json_array(self):
|
||
|
|
assert _try_parse_json('[1, 2, 3]') == [1, 2, 3]
|