File size: 3,411 Bytes
ac9f14b | 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 | """Sanity-check the generated shellops_pro dataset.
For every row:
* re-run `gt_bash` against a fresh copy of init_dir
* confirm stdout matches `expected_text` exactly (for string/hybrid)
* confirm resulting file tree byte-matches post_files (for files/hybrid)
* verify file count, file-type diversity, and total char volume
"""
import shutil
import subprocess
import tempfile
from collections import Counter
from pathlib import Path
import pandas as pd
ROOT = Path(__file__).resolve().parents[3]
DATA_DIR = Path(__file__).resolve().parent
def write_tree(dst: Path, files: list) -> None:
dst.mkdir(parents=True, exist_ok=True)
for entry in files:
p = dst / entry["path"]
p.parent.mkdir(parents=True, exist_ok=True)
p.write_text(entry["content"])
def read_tree(src: Path) -> dict:
result = {}
for p in sorted(src.rglob("*")):
if p.is_file():
result[p.relative_to(src).as_posix()] = p.read_text()
return result
def verify_row(row: dict) -> None:
with tempfile.TemporaryDirectory() as tmp:
work = Path(tmp) / "work"
write_tree(work, list(row["pre_files"]))
r = subprocess.run(
["bash", "-c", row["gt_bash"]],
cwd=work,
capture_output=True,
text=True,
)
assert r.returncode == 0, (
f"{row['id']} gt_bash failed: rc={r.returncode} stderr={r.stderr!r}"
)
got_stdout = r.stdout.rstrip("\n")
if row["task_type"] in ("string", "hybrid"):
assert got_stdout == row["expected_text"], (
f"{row['id']} stdout mismatch:\nexpected={row['expected_text']!r}\n"
f"got={got_stdout!r}"
)
if row["task_type"] in ("files", "hybrid"):
actual_tree = read_tree(work)
expected_tree = {e["path"]: e["content"] for e in row["post_files"]}
only_actual = set(actual_tree) - set(expected_tree)
only_expected = set(expected_tree) - set(actual_tree)
diff_keys = [
k
for k in actual_tree
if k in expected_tree and actual_tree[k] != expected_tree[k]
]
assert actual_tree == expected_tree, (
f"{row['id']} file tree mismatch:\n"
f"only_in_actual={only_actual}\n"
f"only_in_expected={only_expected}\n"
f"content_diffs={diff_keys}"
)
def main() -> None:
df = pd.read_parquet(DATA_DIR / "test.parquet")
assert 1 <= len(df) <= 150, f"unexpected row count: {len(df)}"
print(f"{'id':<24} {'type':<7} {'nfiles':>6} {'nbytes':>9} {'exts':<40}")
for _, row in df.iterrows():
pre = list(row["pre_files"])
n_files = len(pre)
n_bytes = sum(len(e["content"]) for e in pre)
ext_counts = Counter(Path(e["path"]).suffix or "<noext>" for e in pre)
exts = ",".join(sorted(ext_counts))
assert n_files >= 30, f"{row['id']}: n_files={n_files}"
verify_row(row.to_dict())
print(
f"{row['id']:<24} {row['task_type']:<7} {n_files:>6} {n_bytes:>9} {exts:<40}"
)
total_bytes = sum(
sum(len(e["content"]) for e in r["pre_files"]) for _, r in df.iterrows()
)
print(f"\nall {len(df)} tasks verified. total pre_files bytes: {total_bytes:,}")
if __name__ == "__main__":
main()
|