Datasets:
File size: 4,024 Bytes
0338a7c | 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 | #!/usr/bin/env python3
"""Validate question-to-image joins without calling a model or inventing outputs."""
import argparse
import hashlib
import json
import tarfile
from pathlib import Path, PurePosixPath
def rows(path):
return [json.loads(line) for line in path.read_text(encoding="utf-8").splitlines() if line.strip()]
def relative_file(root, relative):
path = PurePosixPath(relative)
if path.is_absolute() or ".." in path.parts or "\\" in relative:
raise ValueError("Unsafe relative file")
candidate = root / relative
candidate.resolve().relative_to(root.resolve())
return candidate
def main():
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--questions", type=Path, required=True)
parser.add_argument("--source-root", type=Path, help="Internal audit only: use original media tree and fallback tar")
parser.add_argument("--limit", type=int)
args = parser.parse_args()
if args.limit is not None and args.limit < 1:
parser.error("limit must be positive")
root = Path(__file__).resolve().parents[1]
manifest_rows = rows(root / "media_manifest.jsonl")
manifest = {row["image_id"]: row for row in manifest_rows}
if len(manifest) != len(manifest_rows):
raise ValueError("Duplicate media IDs")
sources = {}
if args.source_root:
sources = {row["image_id"]: row["source_relative_path"] for row in rows(root / "internal/media_sources.jsonl")}
selected = rows(args.questions)
if len({row["item_id"] for row in selected}) != len(selected):
raise ValueError("Duplicate question IDs")
if args.limit:
selected = selected[:args.limit]
checked = set()
failures = []
archive = None
try:
for question in selected:
domain = {"是", "否"} if question["answer_type"] == "binary" else {"A", "B", "C", "D"}
if question["ground_truth"] not in domain:
raise ValueError("Invalid answer domain")
for part in question["input"]:
if part["type"] != "image_url":
continue
reference = part["image_url"]["url"]
if not reference.startswith("asset://"):
raise ValueError("Unexpected image URI")
image_id = reference[len("asset://"):]
if image_id in checked:
continue
image = manifest[image_id]
checked.add(image_id)
try:
if args.source_root:
if image["status"] == "archived_only":
if archive is None:
archive = tarfile.open(relative_file(args.source_root, "data/non_bmp_image_fallback.tar"))
member = archive.getmember(sources[image_id])
if not member.isfile():
raise ValueError("Non-file archive member")
payload = archive.extractfile(member).read()
else:
payload = relative_file(args.source_root, sources[image_id]).read_bytes()
else:
payload = relative_file(root, image["package_path"]).read_bytes()
if len(payload) != image["bytes"] or hashlib.sha256(payload).hexdigest() != image["sha256"]:
raise ValueError("Image integrity mismatch")
except (OSError, KeyError, ValueError) as error:
failures.append({"image_id": image_id, "error_type": type(error).__name__})
finally:
if archive is not None:
archive.close()
print(json.dumps({"questions_checked": len(selected), "unique_images_checked": len(checked), "failed_images": len(failures), "failures": failures[:20], "model_calls": 0}, ensure_ascii=False, indent=2))
raise SystemExit(1 if failures else 0)
if __name__ == "__main__":
main()
|