test_content_moderation.py

"""Content moderation for Plant / Wildcard ability words."""
import json
import sys
from unittest.mock import patch

import config
import engine
import llm


def _state() -> engine.GameState:
    return engine.GameState()


def test_moderate_flags_via_llm() -> None:
    state = _state()
    with patch.object(llm, "complete", return_value='{"flagged": true}'):
        args = engine._moderate_ability_word_args(
            state, "wildcard_card", {"wildcard_word": "badword"}
        )
    assert args["wildcard_word"] == config.FLAGGED_WORD_REPLACEMENT
    assert args["word_flagged"] is True
    assert state.content_flag_active is True
    assert state.content_flag_ability == "wildcard_card"


def test_moderate_allows_clean_word() -> None:
    state = _state()
    with patch.object(llm, "complete", return_value='{"flagged": false}'):
        args = engine._moderate_ability_word_args(
            state, "plant", {"plant_word": "chaos"}
        )
    assert args["plant_word"] == "chaos"
    assert "word_flagged" not in args
    assert state.content_flag_active is False


def test_commit_replaces_flagged_wildcard_in_pool() -> None:
    state = _state()
    state.waiting_for_ability_choice = True
    state.player_pool = [
        engine.WordEntry(word="legacy", category="noun", tier="power"),
    ]
    with patch.object(llm, "complete", return_value='{"flagged": true}'):
        state = engine.commit_round_ability(
            state, "wildcard_card", {"wildcard_word": "slurword"}
        )
    pool_words = [w.word for w in state.player_pool]
    assert config.FLAGGED_WORD_REPLACEMENT in pool_words
    assert "slurword" not in pool_words
    assert state.pending_ability_args["word_flagged"] is True


def test_forced_plant_uses_replacement() -> None:
    state = _state()
    state.pending_ability_played = "plant"
    state.pending_ability_args = {
        "plant_word": config.FLAGGED_WORD_REPLACEMENT,
        "word_flagged": True,
    }
    assert engine._forced_plant_words(state) == [config.FLAGGED_WORD_REPLACEMENT]


def test_calc_word_score_zeroes_inappropriate() -> None:
    results = [{"word": config.FLAGGED_WORD_REPLACEMENT, "score": 2, "note": "sharp"}]
    total, details = engine.calc_word_score(results, [])
    assert total == 0
    assert details[0]["score"] == 0
    assert details[0]["note"] == "flagged replacement"


def test_sanitize_log_args() -> None:
    out = engine.sanitize_ability_args_for_log(
        {"plant_word": "hidden", "word_flagged": True}
    )
    assert out["plant_word"] == config.FLAGGED_WORD_REPLACEMENT


def main() -> None:
    test_moderate_flags_via_llm()
    test_moderate_allows_clean_word()
    test_commit_replaces_flagged_wildcard_in_pool()
    test_forced_plant_uses_replacement()
    test_calc_word_score_zeroes_inappropriate()
    test_sanitize_log_args()
    print("All content moderation tests passed.")


if __name__ == "__main__":
    main()
    sys.exit(0)