File size: 4,279 Bytes
b6f0739 | 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 | #!/usr/bin/env python3
"""Strict, reproducible checks for the 100 manually reviewed semantic records."""
from __future__ import annotations
import argparse
import json
from collections import Counter
from pathlib import Path
ROOT = Path(__file__).resolve().parents[1]
ANN = ROOT / "annotations"
EXPECTED = {f"CBX100_{index:04d}" for index in range(1, 101)}
def read_jsonl(path: Path) -> list[dict]:
with path.open(encoding="utf-8") as handle:
return [json.loads(line) for line in handle if line.strip()]
def main() -> None:
parser = argparse.ArgumentParser()
parser.add_argument("--integrated", action="store_true", help="also verify metadata/prompt synchronization")
args = parser.parse_args()
rows = read_jsonl(ANN / "visual_semantics.jsonl")
ids = [row.get("sample_id") for row in rows]
errors: list[str] = []
if len(rows) != 100:
errors.append(f"expected 100 rows, found {len(rows)}")
if len(set(ids)) != len(ids):
errors.append("duplicate sample IDs: " + str([item for item, count in Counter(ids).items() if count > 1]))
if set(ids) != EXPECTED:
errors.append(f"ID mismatch; missing={sorted(EXPECTED-set(ids))}, extra={sorted(set(ids)-EXPECTED)}")
required = {
"sample_id", "annotation_method", "primary_form", "visual_features",
"visual_caption_en", "visual_caption_zh", "feature_tree_hint", "confidence",
}
feature_keys = {"openings", "recesses", "protrusions", "repeated_structures", "symmetry", "surface_character"}
for row in rows:
sid = row.get("sample_id", "<missing>")
missing = required - row.keys()
if missing:
errors.append(f"{sid}: missing fields {sorted(missing)}")
continue
if row["annotation_method"] != "manual_visual_inspection_of_four_view_render":
errors.append(f"{sid}: invalid annotation method")
if set(row["visual_features"]) != feature_keys:
errors.append(f"{sid}: visual_features schema mismatch")
if not row["visual_caption_en"].strip() or not row["visual_caption_zh"].strip():
errors.append(f"{sid}: blank caption")
if len(row["feature_tree_hint"]) < 2:
errors.append(f"{sid}: insufficient feature-tree hints")
image = ROOT / "data" / "multiview" / f"{sid}.png"
if not image.is_file():
errors.append(f"{sid}: missing reviewed four-view image")
if len({row["visual_caption_en"] for row in rows}) != 100:
errors.append("English captions are not unique per sample")
if len({row["visual_caption_zh"] for row in rows}) != 100:
errors.append("Chinese captions are not unique per sample")
if args.integrated:
metadata = {row["sample_id"]: row for row in read_jsonl(ANN / "metadata_extended.jsonl")}
prompts = {row["sample_id"]: row for row in read_jsonl(ANN / "prompts.jsonl")}
if set(metadata) != EXPECTED or set(prompts) != EXPECTED:
errors.append("integrated index IDs do not match the 100-sample set")
for visual in rows:
sid = visual["sample_id"]
if sid not in metadata or sid not in prompts:
continue
if metadata[sid]["descriptions"]["en"] != visual["visual_caption_en"]:
errors.append(f"{sid}: metadata English description is not synchronized")
if prompts[sid]["descriptions"]["zh"] != visual["visual_caption_zh"]:
errors.append(f"{sid}: prompt-index Chinese description is not synchronized")
if metadata[sid]["prompts"] != prompts[sid]["prompts"]:
errors.append(f"{sid}: prompt payloads are not synchronized")
if errors:
print("Visual-semantic validation: FAILED")
for error in errors:
print("-", error)
raise SystemExit(1)
print("Visual-semantic validation: PASSED")
print("- 100/100 sample IDs present and unique")
print("- 100/100 four-view images present")
print("- 100/100 English and Chinese captions are sample-specific")
print("- structured features and feature-tree hints complete")
if args.integrated:
print("- metadata and prompt indexes synchronized")
if __name__ == "__main__":
main()
|