Muhammad Faraz Malik
Scaffold bug-fixing agent; implement state schema and Docker sandbox runner
26228ef Download tests/unit/test_state.py from MF180/autonomous-bug-fixing-agent: direct link, hf CLI and curl.
- Browser
- Download file 5.18 kB
-
https://huggingface.co/MF180/autonomous-bug-fixing-agent/resolve/main/tests/unit/test_state.py
- Command line
-
hf download hf://MF180/autonomous-bug-fixing-agent/tests/unit/test_state.py
-
curl -L -o test_state.py https://huggingface.co/MF180/autonomous-bug-fixing-agent/resolve/main/tests/unit/test_state.py
5.18 kB
| """Tests for state models, reducers, and checkpoint serialization.""" | |
| from __future__ import annotations | |
| import json | |
| from pathlib import Path | |
| import pytest | |
| from langchain_core.messages import AIMessage, HumanMessage | |
| from langgraph.checkpoint.memory import MemorySaver | |
| from langgraph.graph import END, START, StateGraph | |
| from pydantic import ValidationError | |
| from agent.state import ( | |
| AgentState, | |
| AgentStatus, | |
| ApprovalDecision, | |
| FilePatch, | |
| FixAttempt, | |
| IssueInput, | |
| RunBudget, | |
| SandboxResult, | |
| initial_state, | |
| ) | |
| def issue(tmp_path: Path) -> IssueInput: | |
| return IssueInput(title="average() is wrong", body="average([2,4,6]) != 4", repo_path=tmp_path) | |
| # ------------------------------------------------------------------------ models | |
| def test_budget_defaults_match_groq_guardrails(): | |
| b = RunBudget() | |
| assert b.max_iterations == 8 | |
| assert b.max_repro_attempts == 2 | |
| def test_budget_caps_iterations_at_eight(n: int): | |
| with pytest.raises(ValidationError): | |
| RunBudget(max_iterations=n) | |
| def test_budget_rejects_tool_output_larger_than_context(): | |
| with pytest.raises(ValidationError): | |
| RunBudget(max_context_tokens=500, max_tool_output_chars=4000) | |
| def test_models_forbid_unknown_fields(tmp_path: Path): | |
| with pytest.raises(ValidationError): | |
| IssueInput(title="t", repo_path=tmp_path, bogus=1) | |
| def test_issue_requires_title(tmp_path: Path): | |
| with pytest.raises(ValidationError): | |
| IssueInput(title="", repo_path=tmp_path) | |
| def test_fix_attempt_iteration_is_one_based(): | |
| with pytest.raises(ValidationError): | |
| FixAttempt(iteration=0) | |
| def test_file_patch_requires_diff(): | |
| with pytest.raises(ValidationError): | |
| FilePatch(path="a.py", unified_diff="") | |
| def test_approval_timestamp_is_timezone_aware(): | |
| assert ApprovalDecision(approved=True).decided_at.tzinfo is not None | |
| def test_agent_status_values_are_strings(): | |
| assert AgentStatus.ESCALATED == "escalated" | |
| def test_sandbox_result_is_the_docker_runner_model(): | |
| from sandbox.docker_runner import SandboxResult as Original | |
| assert SandboxResult is Original | |
| # ------------------------------------------------------------------ initial state | |
| def test_initial_state_is_clean(issue: IssueInput): | |
| s = initial_state(issue) | |
| assert s["iteration"] == 0 and s["repro_attempts"] == 0 | |
| assert s["attempts"] == [] and s["messages"] == [] | |
| assert s["status"] is AgentStatus.INGESTED | |
| assert s["approval"] is None and s["commit_sha"] is None and s["pr_url"] is None | |
| assert s["budget"] == RunBudget() | |
| def test_initial_state_accepts_custom_budget(issue: IssueInput): | |
| assert initial_state(issue, RunBudget(max_iterations=3))["budget"].max_iterations == 3 | |
| def test_models_roundtrip_through_json(issue: IssueInput): | |
| attempt = FixAttempt( | |
| iteration=1, | |
| rationale="divide by len", | |
| patches=[FilePatch(path="calc.py", unified_diff="--- a\n+++ b\n")], | |
| result=SandboxResult(exit_code=1, stdout="F", stderr="", duration_s=0.5), | |
| ) | |
| dumped = json.loads(json.dumps(attempt.model_dump(mode="json"))) | |
| assert FixAttempt.model_validate(dumped) == attempt | |
| issue_dumped = json.loads(issue.model_dump_json()) | |
| assert IssueInput.model_validate(issue_dumped) == issue # Path survives as str | |
| # ---------------------------------------------------------- reducers + checkpointer | |
| def _build_graph(): | |
| def first(state: AgentState) -> dict: | |
| return { | |
| "attempts": [FixAttempt(iteration=1, rationale="first")], | |
| "messages": [AIMessage(content="one", id="m1")], | |
| "iteration": 1, | |
| } | |
| def second(state: AgentState) -> dict: | |
| return { | |
| "attempts": [FixAttempt(iteration=2, rationale="second")], | |
| "messages": [AIMessage(content="two", id="m2")], | |
| "iteration": state["iteration"] + 1, | |
| "status": AgentStatus.FIXING, | |
| } | |
| g = StateGraph(AgentState) | |
| g.add_node("first", first) | |
| g.add_node("second", second) | |
| g.add_edge(START, "first") | |
| g.add_edge("first", "second") | |
| g.add_edge("second", END) | |
| return g | |
| def test_reducers_append_attempts_and_messages(issue: IssueInput): | |
| graph = _build_graph().compile() | |
| out = graph.invoke(initial_state(issue) | {"messages": [HumanMessage(content="hi", id="h")]}) | |
| assert [a.rationale for a in out["attempts"]] == ["first", "second"] | |
| assert [m.content for m in out["messages"]] == ["hi", "one", "two"] | |
| assert out["iteration"] == 2 | |
| assert out["status"] == AgentStatus.FIXING | |
| def test_state_checkpoints_with_memory_saver(issue: IssueInput): | |
| graph = _build_graph().compile(checkpointer=MemorySaver()) | |
| cfg = {"configurable": {"thread_id": "t1"}} | |
| graph.invoke(initial_state(issue), cfg) | |
| snap = graph.get_state(cfg) | |
| assert snap.values["iteration"] == 2 | |
| attempts = snap.values["attempts"] | |
| assert len(attempts) == 2 and all(isinstance(a, FixAttempt) for a in attempts) | |
| assert isinstance(snap.values["issue"], IssueInput) | |
| assert snap.values["issue"].repo_path == issue.repo_path | |