Files
mcp-server/tests/unit/test_result_tool.py
ClaudeandClaude Fable 5 6cf68439a2 feat: add external job tools + improve LLM-facing tool descriptions
Add 3 new MCP tools for inspecting YARN applications NOT submitted
through this service: get_external_job_logs, get_external_job_status,
get_external_job_result. Each takes application_id + connection_name
and queries YARN directly, bypassing the local JobStore.

- spark_executor/tools/external_jobs.py: 3 tool functions
- spark_executor/tools/requests.py: 3 new Pydantic body models
  (ExternalJobLogsRequest, ExternalJobStatusRequest,
  ExternalJobResultRequest)
- spark_executor/server.py: 3 new POST routes with explicit operation_id
- tests/unit/test_external_jobs.py: 7 unit tests
- tests/integration/test_mcp_routes.py: assert 20 tool routes
- README.md: list the 3 new tools

To make the LLM pick the right tool and not guess at field values,
also:

- Add Pydantic field descriptions for 22 fields across 8 request models
  (SaveConnectionRequest, UpdatePendingJobRequest, GetJobLogsRequest,
  JobIdRequest, PendingIdRequest, ConnectionNameRequest, plus the new
  ExternalJob*Request models).
- Update 12 route descriptions with cross-references, prerequisite
  context, and 400 behavior notes.
- Refactor _unknown_job_error: an input that looks like a YARN
  application_id (starts with 'application_') now returns HTTP 400
  (ValueError) with a hint message naming the right external tool;
  other not-found cases still return 404 (KeyError). This catches the
  common LLM mistake of passing application_id to the internal
  get_job_* / kill_job tools.
- 4 new unit tests for the 400 behavior.

Tests: 356 passed (up from 242).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-08 20:03:40 +08:00

194 lines
6.1 KiB
Python

# coding=utf-8
from datetime import datetime
from pathlib import Path
from unittest.mock import patch
import pytest
from spark_executor.core import connection_store
from spark_executor.core.job_store import JobStore
from spark_executor.models import Connection, Job
from spark_executor.tools import result
from spark_executor.tools import connections
@pytest.fixture
def fresh_stores(tmp_path: Path):
"""Wire up connection + job stores rooted in tmp_path. Per-test isolation
so file-backed JobStore doesn't leak between cases."""
store = connection_store.ConnectionStore(data_dir=str(tmp_path))
connection_store.store = store
connections.store = store
result.conn_store = store
result.store = JobStore(data_dir=str(tmp_path))
def _fresh_stores():
store = connection_store.ConnectionStore()
connection_store.store = store
connections.store = store
result.conn_store = store
result.store = JobStore()
def _seed(job_id="abc", app_id="application_1"):
_fresh_stores()
result.conn_store.save(Connection(name="prod", master="yarn", yarn_rm_url="http://rm:8088"))
result.store.put(
Job(
job_id=job_id,
application_id=app_id,
script_path="/tmp/j.py",
queue="default",
submit_time=datetime(2026, 6, 24),
connection="prod",
yarn_rm_url="http://rm:8088",
)
)
def test_result_returns_parsed_fields():
_seed()
raw = {
"app": {
"id": "application_1",
"state": "FINISHED",
"finalStatus": "SUCCEEDED",
"diagnostics": "Application completed successfully",
"trackingUrl": "http://nm:8088/proxy/application_1",
"startedTime": 1700000000000,
"finishedTime": 1700000123000,
}
}
import json
with patch(
"spark_executor.tools.result.get_application_status",
return_value=("FINISHED", json.dumps(raw)),
) as m:
out = result.get_job_result("abc")
assert out.application_id == "application_1"
assert out.state == "FINISHED"
assert out.final_status == "SUCCEEDED"
assert out.diagnostics == "Application completed successfully"
assert out.tracking_url == "http://nm:8088/proxy/application_1"
assert out.started_time == 1700000000000
assert out.finished_time == 1700000123000
args = m.call_args.args
assert args[0] == "application_1"
assert args[1].yarn_rm_url == "http://rm:8088"
def test_result_handles_running_job():
_seed(job_id="running", app_id="application_2")
raw = {
"app": {
"id": "application_2",
"state": "RUNNING",
"finalStatus": "UNDEFINED",
"trackingUrl": "http://rm:8088/proxy/application_2",
"startedTime": 1700000000000,
}
}
import json
with patch(
"spark_executor.tools.result.get_application_status",
return_value=("RUNNING", json.dumps(raw)),
):
out = result.get_job_result("running")
assert out.state == "RUNNING"
assert out.final_status == "UNDEFINED"
assert out.finished_time is None
def test_result_handles_missing_optional_fields():
_seed(job_id="accepted", app_id="application_3")
raw = {"app": {"id": "application_3", "state": "ACCEPTED"}}
import json
with patch(
"spark_executor.tools.result.get_application_status",
return_value=("ACCEPTED", json.dumps(raw)),
):
out = result.get_job_result("accepted")
assert out.state == "ACCEPTED"
assert out.application_id == "application_3"
assert out.final_status is None
assert out.diagnostics is None
assert out.tracking_url is None
assert out.started_time is None
assert out.finished_time is None
def test_result_raises_keyerror_for_unknown_job():
_fresh_stores()
with pytest.raises(KeyError) as ei:
result.get_job_result("missing")
# New error message must still flag "Unknown" so callers / agents can
# recognize the failure, AND mention application_id so the agent
# knows the other form is also accepted.
msg = str(ei.value)
assert "job_id" in msg
assert "application_id" in msg
def test_result_raises_400_for_external_application_id(fresh_stores):
"""Input that looks like a YARN application_id but is not in the local
JobStore must raise ValueError (-> 400) with a hint to use the
external tool, NOT a generic KeyError (-> 404)."""
with pytest.raises(ValueError, match="get_external_job_result"):
result.get_job_result("application_17400000001_0001")
def test_result_raises_when_connection_missing():
_fresh_stores()
result.store.put(
Job(
job_id="abc",
application_id="application_1",
script_path="/tmp/j.py",
queue="default",
submit_time=datetime(2026, 6, 24),
connection="missing",
)
)
with pytest.raises(KeyError, match="Connection not found"):
result.get_job_result("abc")
# --- application_id accepted (regression: "agent passed wrong id" bug) ---
def test_get_job_result_accepts_application_id(fresh_stores):
result.conn_store.save(Connection(name="prod", master="yarn", yarn_rm_url="http://rm:8088"))
result.store.put(
Job(
job_id="a1b2c3d4e5f6",
application_id="application_17400000001_0001",
script_path="/tmp/j.py",
queue="default",
submit_time=datetime(2026, 6, 24),
connection="prod",
yarn_rm_url="http://rm:8088",
)
)
raw = {
"app": {
"id": "application_17400000001_0001",
"state": "SUCCEEDED",
"finalStatus": "SUCCEEDED",
}
}
import json
with patch(
"spark_executor.tools.result.get_application_status",
return_value=("FINISHED", json.dumps(raw)),
) as m:
out = result.get_job_result("application_17400000001_0001")
assert out.application_id == "application_17400000001_0001"
assert out.state == "FINISHED"
assert m.call_args.args[0] == "application_17400000001_0001"