flickr5k / tools /validate_release.py
brianhu25's picture
King Horton
Flickr5k v1.0
49349fe
Raw History Blame Contribute Delete
7.03 kB
#!/usr/bin/env python3
"""Validate a Flickr5k release locally. No credentials, network or ML models needed."""
from __future__ import annotations
import argparse
import csv
import hashlib
import json
import math
import re
from collections import Counter
from pathlib import Path, PurePosixPath
FIELDS = {
"photo_id":"string", "owner_id":"string", "latitude":"float64", "longitude":"float64",
"location_source":"string", "flickr_accuracy":"int32", "camera":"string", "capture_date":"string",
"exif_gps_latitude":"float64", "exif_gps_longitude":"float64", "gps_geotag_distance_m":"float64",
"license_id":"int32", "license_name":"string", "license_url":"string", "license_status":"string",
"license_checked_at":"string", "license_review_category":"string",
"source_url":"string", "review_status":"string", "country":"string", "continent":"string",
"rights_decision":"string", "rights_note":"string", "width":"int32", "height":"int32", "bytes":"int64", "sha256":"string",
}
FIXED = {"README.md","LICENSE_NOTES.md","RELEASE.json","SHA256SUMS","benchmark.csv","ATTRIBUTION.csv",
"RIGHTS_REVIEW.csv","RIGHTS_SUMMARY.json","FILTERING.md","REMOVED.csv","score.py",
".gitattributes",".gitignore","test/metadata.jsonl","tools/validate_release.py",
"tools/audit_dataset.py","tools/audit_semantics.py","tools/audit_report.py"}
def digest(path):
value=hashlib.sha256()
with path.open("rb") as file:
for chunk in iter(lambda:file.read(1024*1024),b""):
value.update(chunk)
return value.hexdigest()
def validate(root,decode_images=False):
root=Path(root).resolve()
paths={p.relative_to(root).as_posix():p for p in root.rglob("*") if p.is_file() or p.is_symlink()}
if not FIXED<=paths.keys():
raise ValueError("Release is missing required files: "+", ".join(sorted(FIXED-paths.keys())))
for relative,path in paths.items():
if path.is_symlink():
raise ValueError("Symlink is not allowed in release: "+relative)
if relative not in FIXED and not re.fullmatch(r"test/images/[0-9]+\.jpg",relative):
raise ValueError("Unexpected/private artifact in release: "+relative)
checksums={}
for line in (root/"SHA256SUMS").read_text().splitlines():
match=re.fullmatch(r"([0-9a-f]{64}) (.+)",line)
if not match or match[2] in checksums:
raise ValueError("Invalid or duplicate checksum entry")
relative=match[2]
if PurePosixPath(relative).is_absolute() or ".." in PurePosixPath(relative).parts:
raise ValueError("Unsafe checksum path")
checksums[relative]=match[1]
if checksums.keys()!=paths.keys()-{"SHA256SUMS"}:
raise ValueError("Checksum manifest does not match release files")
for relative,value in checksums.items():
if digest(paths[relative])!=value:
raise ValueError("Checksum mismatch: "+relative)
rows=[]
with (root/"test/metadata.jsonl").open(encoding="utf-8") as file:
for line in file:
row=json.loads(line)
if set(row)!={"file_name",*FIELDS}:
raise ValueError("Unexpected metadata schema")
for field,dtype in FIELDS.items():
value=row[field]
if value is None:
if field in {"photo_id","owner_id","latitude","longitude","location_source","width","height","sha256"}:
raise ValueError("Required field is null: "+field)
continue
if dtype=="string" and not isinstance(value,str):
raise ValueError("Expected string field: "+field)
if dtype.startswith("int") and (not isinstance(value,int) or isinstance(value,bool)):
raise ValueError("Expected integer field: "+field)
if dtype=="float64" and (not isinstance(value,(int,float)) or isinstance(value,bool) or not math.isfinite(value)):
raise ValueError("Invalid numeric field: "+field)
if not re.fullmatch(r"[0-9]+",row["photo_id"]) or row["file_name"]!=f"images/{row['photo_id']}.jpg":
raise ValueError("Image filename/ID mismatch")
if not -90<=row["latitude"]<=90 or not -180<=row["longitude"]<=180:
raise ValueError("Invalid latitude/longitude")
if not row["source_url"].startswith("https://www.flickr.com/photos/"):
raise ValueError("Invalid source photo page")
relative="test/"+row["file_name"]
if relative not in paths or checksums[relative]!=row["sha256"] or paths[relative].stat().st_size!=row["bytes"]:
raise ValueError("Image/metadata mismatch: "+row["photo_id"])
if min(row["width"],row["height"])<=0 or max(row["width"],row["height"])>1024:
raise ValueError("Invalid saved image dimensions")
if decode_images:
from PIL import Image
with Image.open(paths[relative]) as image:
image.load()
if image.format!="JPEG" or image.size!=(row["width"],row["height"]) or image.getexif():
raise ValueError("Image decode/format/EXIF check failed: "+relative)
rows.append(row)
ids={r["photo_id"] for r in rows}
if len(ids)!=len(rows) or {"test/"+r["file_name"] for r in rows}!={p for p in paths if p.startswith("test/images/")}:
raise ValueError("Duplicate metadata ID or unreferenced image")
release=json.loads((root/"RELEASE.json").read_text())
if release["packaged_images"]!=len(rows):
raise ValueError("Release count mismatch")
if max(Counter(r["owner_id"] for r in rows).values(),default=0)>3:
raise ValueError("Photographer cap exceeded")
with (root/"benchmark.csv").open(newline="") as file:
benchmark=list(csv.DictReader(file))
if len(benchmark)!=len(rows) or {r["IMG_ID"] for r in benchmark}!={r["photo_id"]+".jpg" for r in rows}:
raise ValueError("Benchmark CSV mismatch")
for relative,path in paths.items():
if relative.endswith((".jsonl",".json",".md",".csv",".py")):
text=path.read_text(encoding="utf-8")
if re.search(r"hf_[A-Za-z0-9]{20,}|AKIA[0-9A-Z]{16}|-----BEGIN (?:RSA |EC |OPENSSH )?PRIVATE KEY-----",text):
raise ValueError("Credential-like content in release: "+relative)
return dict(images=len(rows),files=len(paths),labels=dict(Counter(r["location_source"] for r in rows)),
license_status=dict(Counter(r["license_status"] for r in rows)))
def main():
parser=argparse.ArgumentParser(description=__doc__)
parser.add_argument("release",type=Path,nargs="?",default=Path.cwd())
parser.add_argument("--decode-images",action="store_true",help="also decode every image; requires Pillow")
args=parser.parse_args()
print(json.dumps(validate(args.release,args.decode_images),indent=2))
if __name__=="__main__":
main()