GCMD_Keyword_Classifier_MVP / scripts /make_sample_articles.py
igerasimov's picture
Small sclassification script for fake and live runs
f7b61d8
Raw
History Blame Contribute Delete
682 Bytes
# scripts/make_sample_articles.py
import json
from pathlib import Path
source = Path("data/articles.json")
target = Path("data/samples/articles_10.json")
articles = json.loads(source.read_text())
valid = [
a
for a in articles
if isinstance(a, dict)
and isinstance(a.get("DOI"), str)
and a["DOI"]
and isinstance(a.get("Title"), str)
and a["Title"]
and isinstance(a.get("Abstract"), str)
and isinstance(a.get("Year"), int)
and not isinstance(a.get("Year"), bool)
]
target.parent.mkdir(parents=True, exist_ok=True)
target.write_text(json.dumps(valid[:10], indent=2), encoding="utf-8")
print(f"Wrote {len(valid[:10])} articles to {target}")