summarize_votes.py

import argparse
import json
from pathlib import Path
from typing import Dict, List, Optional


def normalize_model_label(label: Optional[str]) -> Optional[str]:
    if label is None:
        return None
    text = label.strip().lower()
    if not text:
        return None
    if text == "base":
        return "base"
    if text == "lora":
        return "lora"
    return text


def summarize_votes(votes: List[Dict]) -> Dict[str, float]:
    base_count = 0
    lora_count = 0
    tie_count = 0
    other_count = 0

    for entry in votes:
        choice = (entry.get("choice") or "").strip().lower()

        display_left = normalize_model_label(
            entry.get("display_left_model") or entry.get("left_model")
        )
        display_right = normalize_model_label(
            entry.get("display_right_model") or entry.get("right_model")
        )

        if choice == "tie":
            tie_count += 1
            continue
        if choice == "left":
            winner = display_left
        elif choice == "right":
            winner = display_right
        else:
            other_count += 1
            continue

        if winner == "base":
            base_count += 1
        elif winner == "lora":
            lora_count += 1
        else:
            other_count += 1

    total = base_count + lora_count + tie_count + other_count
    if total == 0:
        return {
            "base": 0,
            "lora": 0,
            "tie": 0,
            "other": 0,
            "total": 0,
        }

    def pct(value: int) -> float:
        return round((value / total) * 100.0, 2)

    return {
        "base": base_count,
        "base_pct": pct(base_count),
        "lora": lora_count,
        "lora_pct": pct(lora_count),
        "tie": tie_count,
        "tie_pct": pct(tie_count),
        "other": other_count,
        "other_pct": pct(other_count),
        "total": total,
    }


def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser(
        description="Summarize vote results into base/lora/tie counts and percentages.",
    )
    parser.add_argument(
        "--votes",
        type=Path,
        default=Path("votes.json"),
        help="Path to votes JSON file (default: ./votes.json).",
    )
    return parser.parse_args()


def main() -> None:
    args = parse_args()
    votes_path = args.votes

    if not votes_path.exists():
        raise FileNotFoundError(f"Votes file not found: {votes_path}")

    with votes_path.open("r", encoding="utf-8") as f:
        content = f.read().strip()

    if not content:
        print("No votes recorded.")
        return

    data = json.loads(content)
    if not isinstance(data, list):
        raise ValueError("Votes file must contain a JSON array.")

    summary = summarize_votes(data)

    if summary["total"] == 0:
        print("No votes classified.")
        return

    print(f"Total votes: {summary['total']}")
    print(f"Base preference: {summary['base']} ({summary['base_pct']}%)")
    print(f"Lora preference: {summary['lora']} ({summary['lora_pct']}%)")
    print(f"Tie: {summary['tie']} ({summary['tie_pct']}%)")
    if summary["other"]:
        print(f"Unclassified: {summary['other']} ({summary['other_pct']}%)")


if __name__ == "__main__":
    main()