"""Tests for IMP-16-U1 unit u9: ``detect_invented_text``.

Locks the Phase Z port of the deterministic hallucination guard
(Phase Q reference: ``src/content_verifier.py:276-315``). The function
is pure and composes u2 (extract_text_from_html), u3
(normalize_for_comparison), and u4 (extract_keywords). No Phase Q
import is exercised.
"""
from __future__ import annotations

from src.phase_z2_verification_utils import (
    _INVENTED_TEXT_ALLOWED_LABELS,
    _INVENTED_TEXT_CSS_NUMBER_PATTERN,
    _INVENTED_TEXT_KEYWORD_THRESHOLD,
    _INVENTED_TEXT_MIN_LENGTH,
    _INVENTED_TEXT_TRUNCATE_LEN,
    detect_invented_text,
)


def test_detect_invented_text_constants_locked() -> None:
    """Lock the five named module constants ported from Phase Q literals."""
    assert _INVENTED_TEXT_MIN_LENGTH == 15
    assert _INVENTED_TEXT_ALLOWED_LABELS == frozenset(
        {"용어 정의", "핵심 메시지", "상세 비교"}
    )
    assert _INVENTED_TEXT_CSS_NUMBER_PATTERN.pattern == r"^[\d\s.,%px#rgb()]+$"
    assert _INVENTED_TEXT_KEYWORD_THRESHOLD == 0.4
    assert _INVENTED_TEXT_TRUNCATE_LEN == 80


def test_detect_invented_text_returns_empty_when_html_is_in_mdx() -> None:
    """Text whose keywords fully appear in MDX is NOT flagged."""
    mdx = "원본 콘텐츠는 분석에 관한 것입니다."
    html = "<p>원본 콘텐츠는 분석에 관한 것입니다.</p>"
    assert detect_invented_text(mdx, html) == []


def test_detect_invented_text_flags_text_with_low_keyword_overlap() -> None:
    """Text whose keywords do not appear in MDX is flagged as invented."""
    mdx = "원본 콘텐츠는 분석에 관한 것입니다."
    html = "<p>완전히 다른 발명된 텍스트가 여기 있습니다 일반적이지 않은</p>"
    result = detect_invented_text(mdx, html)
    assert len(result) == 1
    assert "발명된" in result[0]


def test_detect_invented_text_skips_short_text() -> None:
    """Text shorter than ``min_length`` is not even considered."""
    mdx = "원본 콘텐츠"
    html = "<p>짧은 텍스트</p>"
    assert detect_invented_text(mdx, html) == []


def test_detect_invented_text_skips_allowed_structural_labels() -> None:
    """Allowed labels are skipped even when keyword overlap is zero.

    Phase Q default ``min_length=15`` makes the allowed-label gate
    unreachable for the bundled labels (all < 15 chars). The Phase Z
    port preserves the gate verbatim — exercised here with
    ``min_length=0`` so the structural-label short-circuit is
    actually observable.
    """
    mdx = "원본 콘텐츠"
    html = "<h2>용어 정의</h2><h2>핵심 메시지</h2><h2>상세 비교</h2>"
    assert detect_invented_text(mdx, html, min_length=0) == []


def test_detect_invented_text_skips_css_number_pattern_fragments() -> None:
    """CSS/numeric fragments (e.g. ``100px 200px 300px``) are skipped."""
    mdx = "원본 콘텐츠"
    html = "<style>.x { padding: 100px; }</style><div>100px 200px 300px</div>"
    assert detect_invented_text(mdx, html) == []


def test_detect_invented_text_truncates_flagged_value_to_80_chars() -> None:
    """A flagged fragment longer than 80 chars is truncated for reporting."""
    mdx = "원본 콘텐츠"
    invented = "발명" * 50
    html = f"<p>{invented}</p>"
    result = detect_invented_text(mdx, html)
    assert len(result) == 1
    assert len(result[0]) == 80
    assert result[0] == invented[:80]