"""Unit tests for the Novice-mode glossary wrap. Pure-function; no DB / HTTP."""
from __future__ import annotations

import pytest

from app.services.glossary import wrap_glossary


def test_no_op_when_tone_is_not_novice():
    """Wrap is gated by tone — INTERMEDIATE and unset both pass through."""
    text = "VIX spiked to 22."
    assert wrap_glossary(text, tone="INTERMEDIATE") == text
    assert wrap_glossary(text, tone=None) == text
    assert wrap_glossary(text, tone="") == text


def test_no_op_when_html_is_empty():
    assert wrap_glossary("", tone="NOVICE") == ""
    assert wrap_glossary(None, tone="NOVICE") == ""


def test_wraps_first_occurrence_only():
    """A term that appears twice gets wrapped only on the first hit —
    repeating tooltips on every word is noisy."""
    out = wrap_glossary("VIX is high; VIX matters.", tone="NOVICE")
    assert out.count('class="glossary"') == 1
    assert '>VIX</span>' in out
    # Second occurrence stays plain.
    assert "; VIX matters" in out


def test_wraps_multiple_distinct_terms():
    out = wrap_glossary("VIX rose; the yield curve flattened.", tone="NOVICE")
    assert 'data-term="VIX"' in out
    assert 'data-term="yield curve"' in out


def test_acronyms_are_case_sensitive():
    """VIX matches; 'vix' alone shouldn't (avoid false positives)."""
    assert 'class="glossary"' in wrap_glossary("VIX up.", tone="NOVICE")
    assert 'class="glossary"' not in wrap_glossary("vix up.", tone="NOVICE")


def test_phrase_terms_match_case_insensitively():
    """'yield curve' should match regardless of capitalisation."""
    out_lower = wrap_glossary("the yield curve flattened.", tone="NOVICE")
    out_title = wrap_glossary("The Yield Curve flattened.", tone="NOVICE")
    assert 'class="glossary"' in out_lower
    assert 'class="glossary"' in out_title


def test_aliases_match():
    """'high-yield OAS' aliases through to the canonical HY OAS entry."""
    out = wrap_glossary("the credit spread widened today.", tone="NOVICE")
    assert 'class="glossary"' in out
    assert 'data-term="HY OAS"' in out


def test_word_boundary_prevents_substring_match():
    """ERP shouldn't match inside 'WERP', 'HERP', etc."""
    out = wrap_glossary("WERPS isn't a term.", tone="NOVICE")
    assert 'class="glossary"' not in out


def test_definition_is_escaped_in_data_attr():
    """A definition with quotes/HTML must be HTML-escaped in attributes
    so it doesn't break the surrounding markup."""
    out = wrap_glossary("VIX moved.", tone="NOVICE")
    # data-def="..." must use &quot; not raw ", &amp; not raw &.
    assert 'data-def="' in out
    # The S&P 500 reference in the VIX definition uses an ampersand; it
    # should be escaped.
    assert "&amp;P 500" in out
    assert '"P 500' not in out  # raw " inside attr would break


def test_skips_content_inside_code_blocks():
    """Wrapping inside <code> would mangle source examples; we skip those."""
    html = "Outside: VIX is up. <code>Inside: VIX is up.</code>"
    out = wrap_glossary(html, tone="NOVICE")
    # The first VIX (outside) should be wrapped.
    assert '<span class="glossary"' in out
    # The VIX inside <code> stays plain.
    assert "<code>Inside: VIX is up.</code>" in out


def test_skips_content_inside_anchor_tags():
    """Wrapping inside <a> would double-up on tooltips and weird the link."""
    html = '<a href="/x">VIX explainer</a> and VIX here too.'
    out = wrap_glossary(html, tone="NOVICE")
    # Anchor content untouched.
    assert '<a href="/x">VIX explainer</a>' in out
    # The non-anchor VIX got wrapped.
    assert '<span class="glossary"' in out


def test_preserves_original_casing_in_wrapped_text():
    """The visible text inside the span should match what was in the source,
    not be replaced with the canonical label."""
    out = wrap_glossary("The Yield Curve is flat.", tone="NOVICE")
    assert ">Yield Curve</span>" in out