Spaces:
Runtime error
Runtime error
File size: 12,509 Bytes
8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 f016eb7 8c486a8 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 | """Tests for TrajectoryLogger -- JSONL export for SFT warm-start."""
import json
from pathlib import Path
import pytest
from open_range.training.trajectory import (
Episode,
TrajectoryLogger,
Turn,
RED_SYSTEM_PROMPT,
BLUE_SYSTEM_PROMPT,
)
# ---------------------------------------------------------------------------
# Turn dataclass
# ---------------------------------------------------------------------------
class TestTurn:
def test_basic_construction(self):
t = Turn(role="red", observation="Range ready.", action="nmap -sV web", reward=0.1)
assert t.role == "red"
assert t.observation == "Range ready."
assert t.action == "nmap -sV web"
assert t.reward == 0.1
assert t.timestamp > 0
def test_timestamp_auto_set(self):
t = Turn(role="blue", observation="ok", action="grep alert", reward=0.0)
assert t.timestamp > 0
# ---------------------------------------------------------------------------
# Episode dataclass
# ---------------------------------------------------------------------------
class TestEpisode:
def _make_episode(self) -> Episode:
ep = Episode(episode_id="ep-1", snapshot_id="snap-1", tier=1)
ep.briefings = {
"red": "Red briefing",
"blue": "Blue briefing",
}
ep.turns = [
Turn(
role="red",
observation="[0.2s] 80/tcp open http",
action="nmap -sV web",
reward=0.1,
assistant_content="<think>\nRecon first.\n</think>",
),
Turn(
role="blue",
observation="[0.3s] suspicious nmap",
action="submit_finding nmap scan",
reward=0.2,
assistant_content="<think>\nThis is actionable.\n</think>",
),
Turn(
role="red",
observation="[0.4s] products",
action="curl http://web/search?q=test",
reward=0.15,
assistant_content="<think>\nInspect the search route.\n</think>",
),
Turn(
role="blue",
observation="[0.5s] SQLi in web log",
action="grep UNION /var/log/siem/web.log",
reward=0.05,
assistant_content="<think>\nI need evidence.\n</think>",
),
Turn(
role="red",
observation="[0.6s] correct",
action="submit_flag FLAG{sqli_123}",
reward=0.5,
assistant_content="<think>\nThis token is worth validating.\n</think>",
tool_name="flag_found",
tool_arguments={"flag": "FLAG{sqli_123}"},
),
]
ep.outcome = "flag_captured"
return ep
def test_red_turns(self):
ep = self._make_episode()
assert len(ep.red_turns) == 3
assert all(t.role == "red" for t in ep.red_turns)
def test_blue_turns(self):
ep = self._make_episode()
assert len(ep.blue_turns) == 2
assert all(t.role == "blue" for t in ep.blue_turns)
def test_total_red_reward(self):
ep = self._make_episode()
expected = 0.1 + 0.15 + 0.5
assert abs(ep.total_red_reward - expected) < 1e-6
def test_total_blue_reward(self):
ep = self._make_episode()
expected = 0.2 + 0.05
assert abs(ep.total_blue_reward - expected) < 1e-6
def test_to_chat_messages_red(self):
ep = self._make_episode()
msgs = ep.to_chat_messages("red")
# system + briefing + 3 * (assistant + tool) = 8 messages
assert len(msgs) == 8
assert msgs[0]["role"] == "system"
assert msgs[0]["content"] == RED_SYSTEM_PROMPT
assert msgs[1]["role"] == "user"
assert msgs[1]["content"] == "Red briefing"
assert msgs[2]["role"] == "assistant"
assert "tool_calls" in msgs[2]
assert msgs[2]["tool_calls"][0]["function"]["name"] == "shell_command"
assert msgs[3]["role"] == "tool"
def test_to_chat_messages_blue(self):
ep = self._make_episode()
msgs = ep.to_chat_messages("blue")
# system + briefing + 2 * (assistant + tool) = 6 messages
assert len(msgs) == 6
assert msgs[0]["role"] == "system"
assert msgs[0]["content"] == BLUE_SYSTEM_PROMPT
assert msgs[1]["role"] == "user"
assert msgs[1]["content"] == "Blue briefing"
def test_to_jsonl_record(self):
ep = self._make_episode()
record = ep.to_jsonl_record("red")
assert record["episode_id"] == "ep-1"
assert record["snapshot_id"] == "snap-1"
assert record["tier"] == 1
assert record["role"] == "red"
assert record["outcome"] == "flag_captured"
assert isinstance(record["messages"], list)
assert isinstance(record["reward"], float)
# Verify it's JSON-serializable
json.dumps(record)
# ---------------------------------------------------------------------------
# TrajectoryLogger
# ---------------------------------------------------------------------------
class TestTrajectoryLogger:
def test_start_episode(self):
logger = TrajectoryLogger()
ep = logger.start_episode(
"ep-1",
snapshot_id="snap-1",
tier=2,
briefings={"red": "brief"},
)
assert ep.episode_id == "ep-1"
assert ep.snapshot_id == "snap-1"
assert ep.tier == 2
assert ep.briefings["red"] == "brief"
assert logger.current_episode is ep
def test_log_turn(self):
logger = TrajectoryLogger()
logger.start_episode("ep-1")
turn = logger.log_turn(role="red", observation="Ready.", action="nmap web", reward=0.1)
assert turn.role == "red"
assert len(logger.current_episode.turns) == 1
def test_log_turn_without_episode_raises(self):
logger = TrajectoryLogger()
with pytest.raises(RuntimeError, match="No active episode"):
logger.log_turn(role="red", observation="x", action="y")
def test_end_episode(self):
logger = TrajectoryLogger()
logger.start_episode("ep-1")
logger.log_turn(role="red", observation="Ready.", action="nmap web", reward=0.1)
ep = logger.end_episode(outcome="timeout", metrics={"steps": 1})
assert ep.outcome == "timeout"
assert ep.metrics == {"steps": 1}
assert logger.current_episode is None
assert len(logger.episodes) == 1
def test_end_episode_without_active_raises(self):
logger = TrajectoryLogger()
with pytest.raises(RuntimeError, match="No active episode"):
logger.end_episode()
def test_start_new_episode_abandons_current(self):
logger = TrajectoryLogger()
logger.start_episode("ep-1")
logger.log_turn(role="red", observation="x", action="y")
logger.start_episode("ep-2")
assert len(logger.episodes) == 1
assert logger.episodes[0].outcome == "abandoned"
assert logger.current_episode.episode_id == "ep-2"
def test_clear(self):
logger = TrajectoryLogger()
logger.start_episode("ep-1")
logger.log_turn(role="red", observation="x", action="y")
logger.end_episode()
logger.clear()
assert len(logger.episodes) == 0
assert logger.current_episode is None
# ---------------------------------------------------------------------------
# JSONL export
# ---------------------------------------------------------------------------
class TestExportJsonl:
def _build_logger_with_episodes(self) -> TrajectoryLogger:
logger = TrajectoryLogger()
# Episode 1: Red succeeds (high reward)
logger.start_episode("ep-1", snapshot_id="snap-1", tier=1)
logger.log_turn(role="red", observation="Range ready.", action="nmap -sV web", reward=0.1)
logger.log_turn(role="blue", observation="Alert: nmap", action="submit_finding nmap", reward=0.2)
logger.log_turn(role="red", observation="80/tcp open", action="curl http://web/search?q=1' OR 1=1--", reward=0.3)
logger.log_turn(role="red", observation="FLAG{sqli}", action="submit_flag FLAG{sqli}", reward=0.5)
logger.end_episode(outcome="flag_captured")
# Episode 2: Both low reward (below typical threshold)
logger.start_episode("ep-2", snapshot_id="snap-2", tier=1)
logger.log_turn(role="red", observation="Range ready.", action="nmap -sV web", reward=0.01)
logger.log_turn(role="blue", observation="No alerts", action="tail /var/log/siem/web.log", reward=0.01)
logger.end_episode(outcome="timeout")
return logger
def test_export_creates_file(self, tmp_path: Path):
logger = self._build_logger_with_episodes()
out = tmp_path / "trajectories.jsonl"
count = logger.export_jsonl(out)
assert out.exists()
assert count > 0
def test_export_all_no_filter(self, tmp_path: Path):
logger = self._build_logger_with_episodes()
out = tmp_path / "all.jsonl"
count = logger.export_jsonl(out, reward_threshold=0.0)
# 2 episodes * 2 roles = 4 lines
assert count == 4
lines = out.read_text().strip().split("\n")
assert len(lines) == 4
def test_export_with_reward_filter(self, tmp_path: Path):
logger = self._build_logger_with_episodes()
out = tmp_path / "filtered.jsonl"
count = logger.export_jsonl(out, reward_threshold=0.1)
# ep-1 red reward = 0.9, ep-1 blue reward = 0.2 -> both pass
# ep-2 red reward = 0.01, ep-2 blue reward = 0.01 -> both filtered
assert count == 2
def test_export_single_role(self, tmp_path: Path):
logger = self._build_logger_with_episodes()
out = tmp_path / "red_only.jsonl"
count = logger.export_jsonl(out, roles=("red",))
# 2 episodes, red only = 2 lines
assert count == 2
def test_export_jsonl_format(self, tmp_path: Path):
logger = self._build_logger_with_episodes()
out = tmp_path / "format.jsonl"
logger.export_jsonl(out)
lines = out.read_text().strip().split("\n")
for line in lines:
record = json.loads(line)
assert "episode_id" in record
assert "snapshot_id" in record
assert "tier" in record
assert "role" in record
assert "messages" in record
assert "reward" in record
assert "outcome" in record
# Messages must follow chat format
msgs = record["messages"]
assert msgs[0]["role"] == "system"
assert msgs[1]["role"] == "user"
for msg in msgs:
assert "role" in msg
assert "content" in msg
for msg in msgs:
if msg["role"] == "assistant":
assert msg["tool_calls"]
def test_export_creates_parent_dirs(self, tmp_path: Path):
logger = self._build_logger_with_episodes()
out = tmp_path / "nested" / "deep" / "trajectories.jsonl"
count = logger.export_jsonl(out)
assert out.exists()
assert count > 0
def test_export_empty_logger(self, tmp_path: Path):
logger = TrajectoryLogger()
out = tmp_path / "empty.jsonl"
count = logger.export_jsonl(out)
assert count == 0
assert out.exists()
assert out.read_text() == ""
def test_red_and_blue_are_independent_examples(self, tmp_path: Path):
"""Red and Blue trajectories produce separate JSONL lines."""
logger = self._build_logger_with_episodes()
out = tmp_path / "independent.jsonl"
logger.export_jsonl(out)
lines = out.read_text().strip().split("\n")
records = [json.loads(line) for line in lines]
# Find ep-1 records
ep1_records = [r for r in records if r["episode_id"] == "ep-1"]
roles = {r["role"] for r in ep1_records}
assert roles == {"red", "blue"}
# Red and Blue have different system prompts
for rec in ep1_records:
system_msg = rec["messages"][0]
if rec["role"] == "red":
assert "penetration tester" in system_msg["content"]
else:
assert "SOC analyst" in system_msg["content"]
|