tests/test_structural_variation.py

"""Voice flips, coordinated-verb splitting, and per-request structure rotation."""

from __future__ import annotations

from app.engine import orchestrator
from app.engine.split import try_split_coordinated_verbs, try_split_long_sentence
from app.engine.variation import rotate
from app.engine.voice import active_to_passive, passive_to_active


def test_passive_to_active_recovers_the_agent():
    text = (
        "Areas for improvement can be identified by businesses that actively "
        "listen to customer feedback."
    )
    assert passive_to_active(text) == (
        "Businesses that actively listen to customer feedback can identify "
        "areas for improvement."
    )


def test_passive_to_active_keeps_perfect_and_past_tense():
    assert passive_to_active("The budget has been approved by the committee.") == (
        "The committee has approved the budget."
    )
    assert passive_to_active("The report was reviewed by readers yesterday.") == (
        "Readers reviewed the report yesterday."
    )


def test_passive_to_active_needs_an_explicit_agent():
    assert passive_to_active("The window was broken.") is None


def test_passive_to_active_skips_negation_needing_do_support():
    # "The editor did not correct mistakes" cannot be built by surface rebuild.
    assert passive_to_active("Mistakes were not corrected by the editor.") is None


def test_active_to_passive_refuses_to_relocate_a_fronted_phrase():
    """Leftovers land after the agent, which would strip the opener's scope."""
    text = (
        "By consistently delivering high-quality service, organizations can "
        "establish a strong reputation."
    )
    assert active_to_passive(text) is None


def test_active_to_passive_refuses_to_strand_a_preverbal_adverb():
    assert active_to_passive(
        "Organizations can also encourage positive recommendations."
    ) is None
    assert active_to_passive("Maria carefully submitted the final report.") is None


def test_active_to_passive_keeps_hyphenated_compounds_intact():
    result = active_to_passive("The team published the high-quality version last week.")
    assert result == "The high-quality version was published by the team last week."


def test_active_and_passive_are_not_mutually_reachable():
    text = "Readers reviewed the report yesterday."
    passive = active_to_passive(text)
    assert passive is not None
    assert passive_to_active(passive) is not None


def test_active_to_passive_refuses_misattached_coordinate_tails():
    """spaCy may hang list verbs as advcl; rebuilding would strand them."""
    text = (
        "Customers appreciate companies that respond quickly to their questions, "
        "resolve issues efficiently, and treat them with respect."
    )
    assert active_to_passive(text) is None


def test_unfront_opener_moves_prepositional_prefix():
    from app.engine.templates import try_unfront_opener

    assert try_unfront_opener(
        "For the success of any business, supplying excellent customer service "
        "is vital."
    ) == (
        "Supplying excellent customer service is vital for the success of any "
        "business."
    )
    assert try_unfront_opener(
        "By consistently delivering high-quality service, organizations can "
        "establish a strong reputation."
    ) == (
        "Organizations can establish a strong reputation by consistently "
        "delivering high-quality service."
    )
    # Discourse connectives must stay fronted.
    assert try_unfront_opener(
        "However, the committee approved the budget today."
    ) is None


def test_coordinated_split_repeats_subject_instead_of_stranding_a_verb():
    text = (
        "By consistently delivering high-quality service, organizations can "
        "establish a strong reputation, gain customer retention, and encourage "
        "positive word-of-mouth recommendations."
    )
    assert try_split_coordinated_verbs(text) == (
        "By consistently delivering high-quality service, organizations can "
        "establish a strong reputation and gain customer retention. "
        "Organizations can also encourage positive word-of-mouth recommendations."
    )


def test_coordinated_split_refuses_ambiguous_relative_clause():
    """The verb list belongs to "organizations", not to "Customers"."""
    text = (
        "Customers appreciate organizations that respond promptly to their "
        "questions, resolve issues efficiently, and treat them with respect."
    )
    assert try_split_coordinated_verbs(text) is None


def test_plain_split_refuses_to_strand_a_bare_verb_phrase():
    text = (
        "By consistently delivering high-quality service, organizations can "
        "establish a strong reputation, gain customer retention, and encourage "
        "positive word-of-mouth recommendations."
    )
    result = try_split_long_sentence(text)
    assert result is None or "Encourage positive" not in result


def test_rotation_keeps_weak_options_behind_the_band():
    options = [("best", 0.90), ("near", 0.78), ("weak", 0.40)]
    for seed in range(6):
        ordered = rotate(
            options, seed=seed, confidence_of=lambda item: item[1]
        )
        assert ordered[-1] == ("weak", 0.40)
        assert set(ordered) == set(options)


def test_rotation_without_seed_preserves_order():
    options = [("a", 0.9), ("b", 0.8)]
    assert rotate(options, seed=None, confidence_of=lambda i: i[1]) == options


def test_same_sentence_takes_different_structures(structural_variation):
    text = "The teacher graded the papers carefully yesterday."
    outputs = {
        orchestrator.rewrite_document(
            text,
            use_lexical_refinement=False,
            use_paraphrase=False,
            use_minilm_safety=False,
        ).text
        for _ in range(12)
    }
    assert len(outputs) > 1
    for candidate in outputs:
        assert "teacher" in candidate
        assert "papers" in candidate


def test_directional_place_is_not_fronted():
    """"To school, Ram went" is marked; locative fronting stays available."""
    outputs = {
        orchestrator.rewrite_document(
            "Ram went to school yesterday happily.",
            use_lexical_refinement=False,
            use_paraphrase=False,
            use_minilm_safety=False,
            variation_seed=seed,
        ).text
        for seed in range(1, 10)
    }
    for candidate in outputs:
        assert not candidate.lower().startswith("to school,")