File size: 17,668 Bytes
a4e4cb2 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 | #!/bin/bash
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
PROJECT_ROOT="${PROJECT_ROOT:-/root/task}"
LOG_DIR="${HARBOR_LOG_DIR:-/logs/verifier}"
PYTHON_BIN="${PYTHON_BIN:-python3}"
mkdir -p "$LOG_DIR"
export SCRIPT_DIR PROJECT_ROOT LOG_DIR
"$PYTHON_BIN" -m pip install --break-system-packages --no-cache-dir pytest==8.3.5 pytest-json-report==1.5.0 PyYAML==6.0.2 >/dev/null 2>&1 || \
"$PYTHON_BIN" -m pip install --no-cache-dir pytest==8.3.5 pytest-json-report==1.5.0 PyYAML==6.0.2 >/dev/null
"$PYTHON_BIN" - <<'PY'
from __future__ import annotations
import importlib.util
import json
import os
import shutil
import subprocess
import sys
import tempfile
from pathlib import Path
SCRIPT_DIR = Path(os.environ["SCRIPT_DIR"]).resolve()
PROJECT_ROOT = Path(os.environ.get("PROJECT_ROOT", "/root/task")).resolve()
LOG_DIR = Path(os.environ.get("LOG_DIR", "/logs/verifier")).resolve()
def _copy_test_tree(src: Path, dst: Path) -> None:
shutil.copytree(src, dst)
test_sh = dst / "test.sh"
if test_sh.exists():
test_sh.unlink()
def _write_runtime_support(base: Path) -> Path:
pkg = base / "verifier_lib"
pkg.mkdir(parents=True, exist_ok=True)
(pkg / "__init__.py").write_text("", encoding="utf-8")
runtime_text = '''from __future__ import annotations
import importlib.util
import json
import subprocess
from pathlib import Path
def emit_report(task_id, public, hidden, notes=None):
return {
"task_id": task_id,
"public": public,
"hidden": hidden,
"notes": notes or [],
}
def print_report(report):
print(json.dumps(report, ensure_ascii=True))
def read_text(path):
return Path(path).read_text(encoding="utf-8")
def run_subprocess(command, cwd):
return subprocess.run(command, cwd=str(cwd), capture_output=True, text=True)
def load_module(name, path):
spec = importlib.util.spec_from_file_location(name, path)
module = importlib.util.module_from_spec(spec)
assert spec and spec.loader
spec.loader.exec_module(module)
return module
def run_checks(section, checks):
results = []
for name, fn in checks:
try:
detail = fn()
passed = detail is not False
error = None
except AssertionError as exc:
detail = None
passed = False
error = str(exc)
except Exception as exc:
detail = None
passed = False
error = f"{type(exc).__name__}: {exc}"
results.append({
"name": name,
"passed": passed,
"detail": detail,
"error": error,
})
return {
"section": section,
"results": results,
"passed": sum(1 for item in results if item["passed"]),
"total": len(results),
}
'''
(pkg / "runtime.py").write_text(runtime_text, encoding="utf-8")
yaml_text = '''from __future__ import annotations
def _parse_scalar(value: str):
if value in {"true", "True"}:
return True
if value in {"false", "False"}:
return False
if value in {"null", "Null", "~"}:
return None
if value.startswith('"') and value.endswith('"'):
return value[1:-1]
if value.startswith("'") and value.endswith("'"):
return value[1:-1]
try:
if any(ch in value for ch in [".", "e", "E"]):
return float(value)
return int(value)
except ValueError:
return value
def safe_load(stream):
text = stream.read() if hasattr(stream, "read") else str(stream)
raw_lines = []
for line in text.splitlines():
if not line.strip() or line.lstrip().startswith("#"):
continue
raw_lines.append(line.rstrip())
index = 0
def parse_block(indent: int):
nonlocal index
mapping = {}
sequence = []
mode = None
while index < len(raw_lines):
line = raw_lines[index]
current_indent = len(line) - len(line.lstrip(" "))
if current_indent < indent:
break
if current_indent > indent:
break
stripped = line.strip()
if stripped.startswith("- "):
mode = mode or "list"
item_text = stripped[2:].strip()
index += 1
if ":" in item_text:
key, value = item_text.split(":", 1)
item = {key.strip(): _parse_scalar(value.strip()) if value.strip() else None}
if index < len(raw_lines):
next_indent = len(raw_lines[index]) - len(raw_lines[index].lstrip(" "))
if next_indent > current_indent:
nested = parse_block(current_indent + 2)
if isinstance(nested, dict):
item.update(nested)
sequence.append(item)
elif item_text:
sequence.append(_parse_scalar(item_text))
else:
sequence.append(parse_block(current_indent + 2))
continue
mode = mode or "dict"
key, value = stripped.split(":", 1)
key = key.strip()
value = value.strip()
index += 1
if value:
mapping[key] = _parse_scalar(value)
else:
mapping[key] = parse_block(current_indent + 2)
return sequence if mode == "list" else mapping
return parse_block(0)
'''
(base / "yaml.py").write_text(yaml_text, encoding="utf-8")
return base
def _run_pytest(target: Path, report_path: Path, cwd: Path, pytest_config: Path | None) -> int:
command = [
sys.executable,
"-m",
"pytest",
str(target),
"-rA",
"-v",
"--json-report",
f"--json-report-file={report_path}",
]
if pytest_config is not None:
command.extend(["-c", str(pytest_config)])
result = subprocess.run(command, cwd=cwd, check=False, env=os.environ.copy())
return result.returncode
def _load_pytest_results(path: Path) -> dict[str, bool]:
data = json.loads(path.read_text(encoding="utf-8"))
return {test["nodeid"]: test["outcome"] == "passed" for test in data.get("tests", [])}
def _generic_pytest_score(outcome_path: Path, process_path: Path) -> dict:
dimensions = []
total_score = 0.0
groups = [("outcome", outcome_path), ("process", process_path)]
present = [(name, path) for name, path in groups if path.exists()]
weight = 100.0 / len(present) if present else 100.0
for name, path in present:
results = _load_pytest_results(path)
total = len(results)
passed = sum(1 for value in results.values() if value)
ratio = passed / total if total else 0.0
score = weight * ratio
total_score += score
dimensions.append(
{
"name": name,
"weight": round(weight, 2),
"tests_matched": total,
"tests_passed": passed,
"scoring": "proportional",
"ratio": round(ratio, 3),
"score": round(score, 2),
}
)
return {
"total_score": round(total_score, 2),
"max_score": 100.0,
"dimensions": dimensions,
}
def _group_to_dimension(name: str, group: dict, weight: float) -> dict:
total = int(group.get("total", 0))
passed = int(group.get("passed", 0))
ratio = (passed / total) if total else 0.0
return {
"name": name,
"weight": round(weight, 2),
"tests_matched": total,
"tests_passed": passed,
"scoring": "proportional",
"ratio": round(ratio, 3),
"score": round(weight * ratio, 2),
}
def _generic_script_score(outcome_report: dict, process_report: dict) -> dict:
groups = []
if outcome_report:
if "public" in outcome_report:
groups.append(("outcome_public", outcome_report["public"]))
if "hidden" in outcome_report:
groups.append(("outcome_hidden", outcome_report["hidden"]))
if process_report:
if "public" in process_report:
groups.append(("process_public", process_report["public"]))
if "hidden" in process_report:
groups.append(("process_hidden", process_report["hidden"]))
groups = [
(name, group)
for name, group in groups
if isinstance(group, dict) and group.get("total", 0) > 0
]
weight = 100.0 / len(groups) if groups else 100.0
dimensions = []
total_score = 0.0
for name, group in groups:
dim = _group_to_dimension(name, group, weight)
total_score += dim["score"]
dimensions.append(dim)
return {
"total_score": round(total_score, 2),
"max_score": 100.0,
"dimensions": dimensions,
}
def _all_script_groups_pass(report: dict) -> bool:
relevant = [report.get("public"), report.get("hidden")]
present = [
group
for group in relevant
if isinstance(group, dict) and group.get("total", 0) > 0
]
return all(group.get("passed", 0) == group.get("total", 0) for group in present)
def _load_module(name: str, path: Path):
spec = importlib.util.spec_from_file_location(name, path)
module = importlib.util.module_from_spec(spec)
assert spec and spec.loader
spec.loader.exec_module(module)
return module
def _run_script_report(name: str, path: Path, support_root: Path) -> dict:
sys.path.insert(0, str(support_root))
module = _load_module(name, path)
if not hasattr(module, "run"):
raise RuntimeError(f"{path} does not expose run()")
return module.run()
def _run_score(score_script: Path, outcome_path: Path, process_path: Path) -> Path:
subprocess.run(
[sys.executable, str(score_script), str(outcome_path), str(process_path)],
cwd=LOG_DIR,
check=True,
env=os.environ.copy(),
)
return LOG_DIR / "score_report.json"
def _compute_score_fallback(score_script: Path, outcome_path: Path, process_path: Path) -> dict:
module = _load_module("legacy_score_module", score_script)
if hasattr(module, "compute_score"):
combined_results = {
**_load_pytest_results(outcome_path),
**_load_pytest_results(process_path),
}
result = module.compute_score(combined_results)
if isinstance(result, tuple) and len(result) == 2:
total, dims = result
return {"total_score": float(total), "max_score": 100.0, "dimensions": dims}
if isinstance(result, dict):
total = result.get("total_score", result.get("total", 0.0))
maximum = result.get("max_score", 100.0)
dims = result.get("dimensions", result.get("dims", []))
return {
"total_score": float(total),
"max_score": float(maximum),
"dimensions": dims,
}
outcome_data = json.loads(outcome_path.read_text(encoding="utf-8"))
process_data = json.loads(process_path.read_text(encoding="utf-8"))
combined_report = {
"tests": outcome_data.get("tests", []) + process_data.get("tests", []),
}
if all(hasattr(module, attr) for attr in ("RUBRIC", "_collect", "_score_dim")):
passed, failed = module._collect(combined_report)
total = 0.0
dimensions = []
for dim in module.RUBRIC:
score = float(module._score_dim(dim, passed, failed))
total += score
dimensions.append(
{
"name": getattr(dim, "name", "dimension"),
"weight": float(getattr(dim, "weight", 0.0)),
"score": round(score, 2),
}
)
return {
"total_score": round(total, 2),
"max_score": 100.0,
"dimensions": dimensions,
}
raise RuntimeError(f"Unsupported score result from {score_script}")
def _write_reward(score_report: dict, outcome_code: int, process_code: int) -> None:
total = float(score_report.get("total_score", 0.0))
maximum = float(score_report.get("max_score", 100.0)) or 100.0
normalized = total / maximum
payload = {
"total_score": total,
"max_score": maximum,
"normalized_score": normalized,
"outcome_passed": 1.0 if outcome_code == 0 else 0.0,
"process_passed": 1.0 if process_code == 0 else 0.0,
}
LOG_DIR.mkdir(parents=True, exist_ok=True)
(LOG_DIR / "reward.txt").write_text(f"{normalized:.6f}\n", encoding="utf-8")
(LOG_DIR / "reward.json").write_text(
json.dumps(payload, indent=2) + "\n", encoding="utf-8"
)
if not PROJECT_ROOT.exists():
raise FileNotFoundError(f"project root not found: {PROJECT_ROOT}")
with tempfile.TemporaryDirectory(prefix="harbor-eval-") as tmpdir:
runtime_root = Path(tmpdir)
runtime_task_root = runtime_root / "workspace" / "task"
project_root = runtime_task_root / "project"
runtime_test_root = runtime_task_root / "test"
support_root = runtime_root / "support"
runtime_task_root.mkdir(parents=True)
shutil.copytree(PROJECT_ROOT, project_root)
_copy_test_tree(SCRIPT_DIR, runtime_test_root)
task_root = SCRIPT_DIR.parent
instruction_path = task_root / "instruction.md"
if instruction_path.exists():
shutil.copy2(instruction_path, runtime_task_root / "instruction.md")
_write_runtime_support(support_root)
os.environ["PYTHONPATH"] = str(support_root) + os.pathsep + os.environ.get("PYTHONPATH", "")
os.environ["npm_config_cache"] = str(runtime_root / "npm-cache")
outcome_path = runtime_test_root / "test_outcome.py"
process_path = runtime_test_root / "test_process.py"
score_path = runtime_test_root / "score.py"
pytest_config = runtime_test_root / "pytest.ini"
pytest_config = pytest_config if pytest_config.exists() else None
uses_script_runtime = False
for candidate in (outcome_path, process_path):
if candidate.exists() and "verifier_lib.runtime" in candidate.read_text(
encoding="utf-8", errors="ignore"
):
uses_script_runtime = True
break
if uses_script_runtime:
outcome_report = (
_run_script_report("harbor_outcome", outcome_path, support_root)
if outcome_path.exists()
else {}
)
process_report = (
_run_script_report("harbor_process", process_path, support_root)
if process_path.exists()
else {}
)
LOG_DIR.mkdir(parents=True, exist_ok=True)
(LOG_DIR / "outcome_report.json").write_text(
json.dumps(outcome_report, indent=2) + "\n", encoding="utf-8"
)
(LOG_DIR / "process_report.json").write_text(
json.dumps(process_report, indent=2) + "\n", encoding="utf-8"
)
outcome_code = 0 if _all_script_groups_pass(outcome_report) else 1
process_code = 0 if _all_script_groups_pass(process_report) else 1
score_report = _generic_script_score(outcome_report, process_report)
(LOG_DIR / "score_report.json").write_text(
json.dumps(score_report, indent=2) + "\n", encoding="utf-8"
)
_write_reward(score_report, outcome_code, process_code)
raise SystemExit(0 if outcome_code == 0 and process_code == 0 else 1)
outcome_json = runtime_root / "outcome.json"
process_json = runtime_root / "process.json"
outcome_code = _run_pytest(outcome_path, outcome_json, runtime_task_root, pytest_config)
process_code = _run_pytest(process_path, process_json, runtime_task_root, pytest_config)
if score_path.exists():
try:
score_report_path = _run_score(score_path, outcome_json, process_json)
except subprocess.CalledProcessError:
score_report = _compute_score_fallback(score_path, outcome_json, process_json)
score_report_path = LOG_DIR / "score_report.json"
score_report_path.write_text(
json.dumps(score_report, indent=2) + "\n", encoding="utf-8"
)
if score_report_path.exists():
score_report = json.loads(score_report_path.read_text(encoding="utf-8"))
else:
score_report = _compute_score_fallback(score_path, outcome_json, process_json)
score_report_path.write_text(
json.dumps(score_report, indent=2) + "\n", encoding="utf-8"
)
else:
score_report = _generic_pytest_score(outcome_json, process_json)
LOG_DIR.mkdir(parents=True, exist_ok=True)
(LOG_DIR / "score_report.json").write_text(
json.dumps(score_report, indent=2) + "\n", encoding="utf-8"
)
_write_reward(score_report, outcome_code, process_code)
raise SystemExit(0 if outcome_code == 0 and process_code == 0 else 1)
PY
if [ ! -f "$LOG_DIR/reward.txt" ]; then
echo 0 > "$LOG_DIR/reward.txt"
fi
exit 0
|