# Copyright 2026 Umbra Tools (umbratools.com)
# SPDX-License-Identifier: Apache-2.0

"""Guard tests for the character-discovery prompt's over-detection rules.

These rules exist because discovery was missing obvious speakers (characters
identified only by a comparative/ordinal designation, e.g. "the tallest",
"the oldest" in a crowd of six) and producing polluted names ("Walking tallest").
If someone rewrites CHARACTER_DISCOVERY_PROMPT.md and drops them, these fail.
"""

from __future__ import annotations

from pathlib import Path

import pytest

from umbraaudio.analysis.character_discovery import (
    PROMPT_PATH,
    CharacterDiscoverer,
)


@pytest.fixture(scope="module")
def prompt() -> str:
    d = CharacterDiscoverer.__new__(CharacterDiscoverer)
    d._prompt_template = None
    return d._load_prompt()


def test_prompt_loads_and_renders(prompt: str) -> None:
    # The real code path uses .replace(), not .format(), because the JSON schema
    # block contains literal braces.
    rendered = prompt.replace("{known_speakers_section}", "(Aksenzo, Hallr)")
    assert "{known_speakers_section}" not in rendered
    assert len(rendered) > 5000


def test_rule_8_every_speaking_line_is_a_speaker(prompt: str) -> None:
    assert "Rule 8: Every Speaking Line Is a Speaker" in prompt
    # Must name comparative/ordinal designations explicitly.
    for designation in ["the tallest", "the shortest", "the youngest", "the oldest"]:
        assert designation in prompt, f"missing designation example: {designation}"
    # Must state the over-detection bias.
    low = prompt.lower()
    assert "still list both" in low, "must instruct to list both when unsure"


def test_rule_9_name_hygiene(prompt: str) -> None:
    assert "Rule 9: Name Hygiene" in prompt
    low = prompt.lower()
    assert "narration verbs" in low
    # The concrete bug this rule fixes.
    assert "Walking tallest" in prompt
    # Internal labels must never end up in a name.
    assert "(Scene 2)" in prompt


def test_unidentified_covers_designations(prompt: str) -> None:
    low = prompt.lower()
    assert "designation" in low
    # Must forbid upgrading a designation into a guessed proper name.
    assert "never becomes" in low or "not upgrade" in low


def test_prompt_file_is_markdown() -> None:
    assert PROMPT_PATH.suffix == ".md"
    assert PROMPT_PATH.exists()