"""SO-arm datasets in the listing, and the per-author sample that was read. usage: sample.py OUT_DIR/datasets.jsonl OUT_DIR Replaces OUT_DIR/so_ids.txt (every SO-arm id, sorted) and OUT_DIR/sample_ids.txt (up to 3 per author, sorted). SO-arm rule: the case-insensitive pattern below matches anywhere in the full repository id (author namespace included) or in any tag. Sample: for each author in sorted order, that author's sorted ids are shuffled with one random.Random(SEED) and the first 3 kept.""" import collections, json, os, random, re, sys SO = re.compile(r"so[\W_]?10[01]|so[\W_]?arm|so[\W_]?follower", re.I) SEED = 20261118 rows = [json.loads(l) for l in open(sys.argv[1]) if l.strip()] ids = sorted({r["id"] for r in rows if r.get("id") and SO.search(" ".join([r["id"]] + (r.get("tags") or [])))}) by = collections.defaultdict(list) for i in ids: by[i.split("/")[0]].append(i) rng, sample = random.Random(SEED), [] for a in sorted(by): xs = sorted(by[a]); rng.shuffle(xs); sample += xs[:3] os.makedirs(sys.argv[2], exist_ok=True) for name, xs in (("so_ids.txt", ids), ("sample_ids.txt", sorted(sample))): with open(os.path.join(sys.argv[2], name), "w") as f: f.write("".join(x + "\n" for x in xs)) print(len(rows), "listed;", len(ids), "SO-arm from", len(by), "authors;", len(sample), "sampled")