umm / scripts /make_portable_data.py
Orangerl's picture
Add files using upload-large-folder tool
20b4034 verified
Raw
History Blame Contribute Delete
1.97 kB
#!/usr/bin/env python3
from __future__ import annotations
import argparse
import os
from pathlib import Path
import ujson
def main() -> None:
parser = argparse.ArgumentParser(description="Rewrite CoVT JSON image paths to portable basenames.")
parser.add_argument("source_json", type=Path)
parser.add_argument("image_dir", type=Path)
parser.add_argument("output_json", type=Path)
args = parser.parse_args()
args.output_json.parent.mkdir(parents=True, exist_ok=True)
count = 0
missing = 0
first = True
with args.source_json.open("r", encoding="utf-8") as src, args.output_json.open(
"w", encoding="utf-8"
) as dst:
dst.write("[\n")
for raw in src:
line = raw.strip()
if not line or line in ("[", "]"):
continue
if line.endswith(","):
line = line[:-1]
item = ujson.loads(line)
images = item.get("image")
if isinstance(images, str):
basename = os.path.basename(images)
item["image"] = basename
missing += int(not (args.image_dir / basename).is_file())
elif isinstance(images, list):
basenames = [os.path.basename(value) for value in images]
item["image"] = basenames
missing += sum(not (args.image_dir / value).is_file() for value in basenames)
else:
raise TypeError(f"row {count} has invalid image field: {type(images).__name__}")
if not first:
dst.write(",\n")
dst.write(" " + ujson.dumps(item, ensure_ascii=False))
first = False
count += 1
dst.write("\n]\n")
if missing:
raise FileNotFoundError(f"{missing} image references were not found")
print(f"portable_rows={count} missing_images={missing} output={args.output_json}")
if __name__ == "__main__":
main()