"""Build the frozen certificate eligibility snapshot from judge verdicts.

Run this explicitly when the organizers want to refresh ``eligible.json``.
The certificate Space itself remains independent of the verdicts dataset.
"""

import json
import urllib.request
from collections import defaultdict
from pathlib import Path


VERDICTS_URL = (
    "https://huggingface.co/datasets/ICML-2026-agent-repro/verdicts/"
    "resolve/main/verdicts.json"
)
POINTS = {"verified": 2, "falsified": 2, "toy": 1}


def build_eligible(verdicts):
    records = defaultdict(
        lambda: {
            "logbooks": 0,
            "scored_logbooks": 0,
            "scored_claims": 0,
            "points": 0,
        }
    )
    for space_id, verdict in verdicts.items():
        username = space_id.split("/", 1)[0].lower()
        claim_points = [
            POINTS.get(str(claim.get("verdict", "")).lower(), 0)
            for claim in verdict.get("claims", [])
        ]
        rec = records[username]
        rec["logbooks"] += 1
        if any(claim_points):
            rec["scored_logbooks"] += 1
            rec["scored_claims"] += sum(point > 0 for point in claim_points)
            rec["points"] += sum(claim_points)

    return {
        username: rec
        for username, rec in sorted(records.items())
        if rec["points"] > 0
    }


def main():
    with urllib.request.urlopen(VERDICTS_URL) as response:
        verdicts = json.load(response)
    destination = Path(__file__).with_name("eligible.json")
    destination.write_text(
        json.dumps(build_eligible(verdicts), indent=2) + "\n", encoding="utf-8"
    )
    print(f"Wrote {len(build_eligible(verdicts))} eligible participants to {destination}")


if __name__ == "__main__":
    main()