"""Phase A guards: shape, invention, no dropped units."""

from __future__ import annotations

import sys
from pathlib import Path

ROOT = Path(__file__).resolve().parent.parent
sys.path.insert(0, str(ROOT))

from app.pipeline.alignment import iter_source_units
from app.pipeline.candidate_validator import validate_candidate
from app.pipeline.grammar_fix import correct_text
from app.pipeline.orchestrator import _accept_lm_unit, _looks_needs_lm, rewrite_text


def test_rejects_multi_sentence_invention() -> None:
    src = (
        "Eating fast foods is becoming very common and people don't realize "
        "how much it affects their health."
    )
    bad = (
        "Fast food is becoming increasingly prevalent, and many individuals aren't "
        "aware of its detrimental impact on their health due to a lack of awareness "
        "and education."
    )
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "invention" in v.reasons or "shape" in v.reasons or "length" in v.reasons
    ok, reasons = _accept_lm_unit(src, bad)
    assert not ok
    print("multi-sentence/invention reject OK:", v.reasons, reasons)


def test_rejects_two_sentence_for_one() -> None:
    src = "Exercise is one of the most important activities that keeps our body fit."
    bad = (
        "In addition to our exercise routines, keeping our bodies fit is also crucial. "
        "Exercise remains essential."
    )
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "shape" in v.reasons
    print("shape reject OK")


def test_skip_short_and_light() -> None:
    assert not _looks_needs_lm("Yes it is.", 1)  # <4 words
    assert not _looks_needs_lm(
        "Nowadays many people are living an unhealthy life because they don't have enough time.",
        0,
    )
    assert _looks_needs_lm(
        "Nowadays many people are living an unhealthy life because they don't have enough time.",
        1,
    )
    # Short claims must still go to LM at Normal (was ≥8 — grammar-only bug)
    assert _looks_needs_lm("Another important thing is sleeping.", 1)
    assert _looks_needs_lm("Without enough sleep, the brain works poorly.", 1)
    print("skip heuristics OK")


def test_sleep_style_must_visibly_rewrite() -> None:
    """User expectation: not grammar-only; wording must clearly change."""
    import random
    from unittest.mock import patch

    from app.pipeline.orchestrator import (
        _is_true_rewrite,
        _require_visible_rewrite,
        _similarity_ratio,
        _stem_jaccard,
    )

    rng = random.Random(7)
    src = "Another important thing is sleeping."
    out = _require_visible_rewrite(src, src, "Neutral", rng, target_sim=0.85)
    assert _is_true_rewrite(src, out), out
    assert _similarity_ratio(src, out) < 0.90
    assert _stem_jaccard(src, out) < 0.92
    assert "sleep" in out.lower()
    print("sleep opener rewrite OK:", out)

    src2 = "Without enough sleep, the brain."
    out2 = _require_visible_rewrite(src2, src2, "Neutral", rng, target_sim=0.85)
    assert _is_true_rewrite(src2, out2), out2
    assert "sleep" in out2.lower()
    print("without-enough rewrite OK:", out2)

    raw = "Another important thing are sleeping. Without enough sleep, the brain"
    g = correct_text(raw)
    with patch(
        "app.pipeline.orchestrator.rewrite_unit",
        side_effect=lambda u, **_k: u,
    ):
        from app.pipeline.orchestrator import _hybrid_rewrite

        out, _stats, _ = _hybrid_rewrite(
            g, "Neutral", 1, rng, can_generate=True
        )
    assert _similarity_ratio(g, out) < 0.90
    assert "sleep" in out.lower()
    # Must not be grammar-only near-copy of the classic paste
    assert "another important thing is sleeping" not in out.lower()
    print("sleep document hybrid rewrite OK:", out)


def test_exercise_tv_must_not_be_reorder_only() -> None:
    """Exercise/TV sample should change via reorder and/or wording; keep claims."""
    import random
    from unittest.mock import patch

    from app.pipeline.orchestrator import (
        _hybrid_rewrite,
        _is_true_rewrite,
        _similarity_ratio,
    )

    raw = (
        "Exercise is one of the most important activity that keeps our body fit. "
        "If a person exercise regularly, they will have less chance to gets diseases. "
        "Unfortunately, many peoples prefers watching television instead of doing "
        "physical activities."
    )
    g = correct_text(raw)
    rng = random.Random(11)
    with patch(
        "app.pipeline.orchestrator.rewrite_unit",
        side_effect=lambda u, **_k: u,
    ):
        out, stats, _ = _hybrid_rewrite(g, "Neutral", 1, rng, can_generate=True)
    low = out.lower()
    assert out.strip() != g.strip()
    assert "tv" in low or "television" in low
    assert "fit" in low or "exercise" in low or "work out" in low or "workout" in low
    from app.pipeline.alignment import iter_source_units

    src_units = [u for _i, u in iter_source_units(g)]
    out_units = [u for _i, u in iter_source_units(out)]
    assert len(out_units) == len(src_units)
    true_n = sum(1 for s, o in zip(src_units, out_units) if _is_true_rewrite(s, o))
    assert true_n >= max(1, len(src_units) - 1), (true_n, out)
    print("exercise/TV rewrite OK:", out)
    print("  stats", stats)
    _ = _similarity_ratio  # kept for debugging imports


def test_complete_wording_rewrite_not_shuffle() -> None:
    """User expects structural reorder, e.g. time/manner fronting."""
    import random

    from app.pipeline.orchestrator import (
        _is_structural_reorder,
        _is_true_rewrite,
        _require_visible_rewrite,
        _sentence_reorder_variants,
        _try_restructure_unit,
    )
    from app.pipeline.restructure import restructure_sentence

    src = "Ram went to school yesterday happily."
    expected_style = "Yesterday, Ram happily went to school."
    assert _is_true_rewrite(src, expected_style), "reorder must count as rewrite"
    assert _is_structural_reorder(src, expected_style)

    rr = restructure_sentence(src)
    assert rr is not None, "restructure_sentence must handle Ram"
    assert _is_structural_reorder(src, rr.text), rr.text
    assert "yesterday" in rr.text.lower()
    assert "happily" in rr.text.lower()
    assert "headed" not in rr.text.lower()
    assert "cheerfully" not in rr.text.lower()
    assert "day before" not in rr.text.lower()

    try_out = _try_restructure_unit(src)
    assert try_out is not None
    assert _is_structural_reorder(src, try_out), try_out

    variants = _sentence_reorder_variants(src)
    assert any("yesterday, ram happily went to school" in v.lower() for v in variants), variants

    out = _require_visible_rewrite(
        src, src, "Neutral", random.Random(5), target_sim=0.95
    )
    assert _is_true_rewrite(src, out), out
    assert _is_structural_reorder(src, out), out
    assert "ram" in out.lower() and "school" in out.lower()
    assert "yesterday" in out.lower()
    assert "happily" in out.lower()
    assert "went" in out.lower()
    # Must not synonym-thrash the Ram sentence
    assert "headed" not in out.lower()
    assert "cheerfully" not in out.lower()
    assert "day before" not in out.lower()
    print("Ram sentence reorder OK:", out)


def test_restructure_time_only_and_manner() -> None:
    """Additional reorder templates: time front; manner before verb."""
    from app.pipeline.orchestrator import (
        _is_structural_reorder,
        _try_restructure_unit,
    )
    from app.pipeline.restructure import restructure_sentence

    # Time-only → front time
    src_time = "She finished the report yesterday."
    rr_time = restructure_sentence(src_time)
    assert rr_time is not None, "time-only restructure"
    assert rr_time.text.lower().startswith("yesterday"), rr_time.text
    assert _is_structural_reorder(src_time, rr_time.text) or "yesterday" in rr_time.text.lower()
    assert "finished" in rr_time.text.lower() and "report" in rr_time.text.lower()
    print("time-only reorder OK:", rr_time.text)

    # Manner + place → manner before verb
    src_manner = "They walked to the park slowly."
    out_m = _try_restructure_unit(src_manner) or (
        restructure_sentence(src_manner).text if restructure_sentence(src_manner) else None
    )
    assert out_m is not None, "manner reorder"
    low = out_m.lower()
    assert "slowly" in low and "park" in low and "walked" in low
    # Manner should appear before the verb when template applies
    if "slowly walked" in low or low.index("slowly") < low.index("walked"):
        pass
    else:
        # Accept time/place template variants as long as words preserved
        assert _is_structural_reorder(src_manner, out_m) or set(
            w for w in low.replace(".", "").split() if len(w) > 2
        ) >= {"they", "walked", "park", "slowly"}
    print("manner reorder OK:", out_m)

    # Subject manner verb place style
    src_happy = "Maya drove to work quickly."
    rr_h = restructure_sentence(src_happy)
    assert rr_h is not None
    assert "maya" in rr_h.text.lower() and "quickly" in rr_h.text.lower()
    assert "work" in rr_h.text.lower()
    print("manner+place reorder OK:", rr_h.text)


def test_require_visible_keeps_structural_reorder() -> None:
    """Existing structural reorder must not be synonym-polished away."""
    import random

    from app.pipeline.orchestrator import (
        _is_structural_reorder,
        _require_visible_rewrite,
    )

    src = "Ram went to school yesterday happily."
    cand = "Yesterday, Ram happily went to school."
    out = _require_visible_rewrite(
        src, cand, "Neutral", random.Random(1), target_sim=0.5
    )
    assert _is_structural_reorder(src, out), out
    assert "headed" not in out.lower()
    assert "cheerfully" not in out.lower()
    print("keep structural reorder OK:", out)


def test_rejects_degree_adv_junk_reorder() -> None:
    """Do not ship 'more makes' style degree-adverb reorders."""
    from app.pipeline.orchestrator import _try_restructure_unit
    from app.pipeline.restructure import restructure_sentence

    src = "Online learning makes education more accessible for students in remote areas."
    rr = restructure_sentence(src)
    if rr is not None:
        assert "more makes" not in rr.text.lower(), rr.text
    try_out = _try_restructure_unit(src)
    if try_out is not None:
        assert "more makes" not in try_out.lower(), try_out
    print("degree-adv junk reorder rejected OK:", rr, try_out)


def test_because_front_preserves_clauses() -> None:
    from app.pipeline.restructure import restructure_sentence

    src = (
        "Nowadays many people are living an unhealthy life "
        "because they don't have enough time."
    )
    rr = restructure_sentence(src)
    assert rr is not None, "because_front should apply"
    assert rr.template_id == "because_front"
    low = rr.text.lower()
    assert low.startswith("because")
    assert "unhealthy" in low and "enough time" in low
    assert "nowadays" in low
    print("because_front OK:", rr.text)


def test_esl_wording_avoids_synonym_thrash() -> None:
    import random

    from app.pipeline.orchestrator import _require_visible_rewrite

    sleep = "Sleeping is still an important thing."
    out_s = _require_visible_rewrite(
        sleep, sleep, "Neutral", random.Random(3), target_sim=0.95
    )
    low_s = out_s.lower()
    assert "endeavor" not in low_s and "nevertheless" not in low_s
    assert "sleep" in low_s or "sleeping" in low_s
    print("sleep wording OK:", out_s)

    climate = "Climate change is causing more extreme weather events around the world."
    out_c = _require_visible_rewrite(
        climate, climate, "Neutral", random.Random(7), target_sim=0.95
    )
    low_c = out_c.lower()
    assert "uttermost" not in low_c and "upwind" not in low_c
    assert "climate" in low_c and "weather" in low_c
    print("climate wording OK:", out_c)


def test_rejects_hollow_ons_bills() -> None:
    src = (
        "People should save money for emergencies instead of spending every paycheck."
    )
    bad = (
        "People should save money for emergencies instead of spending it ONS bills or bills."
    )
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "broken" in v.reasons or "key_content" in v.reasons
    print("hollow ONS bills reject OK:", v.reasons)


def test_assembly_preserves_unit_count() -> None:
    raw = (
        "Nowadays many peoples are living unhealthy life because they don't have enough times. "
        "Eating fast foods are becoming very common and peoples don't realizes how much it affect their health.\n\n"
        "Exercise is one of the most important activity that keeps our body fit. "
        "If a person exercise regularly, they will have less chance to gets diseases. "
        "Unfortunately, many peoples prefers watching television instead of doing physical activities"
    )
    source = correct_text(raw)
    n_in = len(iter_source_units(source))
    r = rewrite_text(raw, tone="Neutral", strength=1, ml_polish=False)
    # Classical path still grammar-fixes; count sentences in output vs source units
    n_out = len(iter_source_units(r.text))
    assert n_out >= n_in - 1  # allow minor splitter variance after grammar
    assert "television" in r.text.lower() or "physical" in r.text.lower()
    print("unit preservation OK in=", n_in, "out=", n_out)


def test_live_bug_patterns_rejected() -> None:
    src = (
        "Nowadays many people are living an unhealthy life because they don't have enough time."
    )
    invented = (
        "Many people these days live unfulfilling unhealthy lives due to lack of "
        "awareness and education."
    )
    v = validate_candidate(src, invented, min_meaning=0.5)
    assert not v.ok
    assert "invention" in v.reasons
    print("live-bug invention OK")


def test_rejects_fast_food_collocation_drop() -> None:
    src = (
        "Eating fast foods is becoming very common and people don't realize "
        "how much it affects their health."
    )
    # FLAN slip: drops "fast", keeps bare "Food"
    bad = (
        "Food is becoming very common, and people don't realize how much it "
        "affects their health."
    )
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "collocation" in v.reasons
    ok, reasons = _accept_lm_unit(src, bad)
    assert not ok
    print("fast-food collocation reject OK:", v.reasons, reasons)


def test_rejects_workout_program_invention() -> None:
    src = (
        "Unfortunately, many people prefer watching television instead of "
        "doing physical activities."
    )
    bad = (
        "Unfortunately, many people prefer watching television instead of "
        "following a workout program."
    )
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "invention" in v.reasons
    ok, reasons = _accept_lm_unit(src, bad)
    assert not ok
    print("workout-program invention reject OK:", v.reasons, reasons)


def test_keeps_fast_food_when_preserved() -> None:
    src = (
        "Eating fast foods is becoming very common and people don't realize "
        "how much it affects their health."
    )
    good = (
        "Eating fast food is getting very common, and people don't realize "
        "how much it harms their health."
    )
    from app.pipeline.candidate_validator import _protected_collocations_ok

    assert _protected_collocations_ok(src, good)
    print("fast-food keep OK")


def test_rejects_live_mangled_online_learning() -> None:
    src = (
        "Many students prefer to study from home because it saves time and money."
    )
    bad = (
        "It takes less time to study from home than it would at home."
    )
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert any(r in v.reasons for r in ("key_content", "broken", "coverage", "drift"))
    ok, reasons = _accept_lm_unit(src, bad)
    assert not ok
    print("online-learning mangle reject OK:", v.reasons, reasons)


def test_rejects_dropped_online_modifier() -> None:
    src = "Online learning has become more popular during recent years."
    bad = "Learning is becoming more popular in recent years."
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "key_content" in v.reasons
    print("online modifier reject OK:", v.reasons)


def test_rejects_mangled_climate_claim() -> None:
    src = (
        "Climate change is a serious problem that affects every country."
    )
    bad = "The impact of climate change has every country."
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert any(r in v.reasons for r in ("key_content", "coverage", "drift"))
    print("climate mangle reject OK:", v.reasons)


def test_accepts_faithful_paraphrase() -> None:
    src = (
        "Many students prefer to study from home because it saves time and money."
    )
    good = (
        "A lot of students prefer studying at home since it saves them time and money."
    )
    from app.pipeline.candidate_validator import _key_content_ok, _broken_rewrite_ok

    assert _key_content_ok(src, good)
    assert _broken_rewrite_ok(src, good)
    print("faithful paraphrase key-content OK")


def test_rejects_dropped_books_and_cities() -> None:
    src = "Reading books regularly improves vocabulary and critical thinking."
    bad = "Reading helps improve vocabulary and critical thinking."
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "key_content" in v.reasons
    print("books drop reject OK:", v.reasons)

    src2 = "Public transport is cheaper than owning a car in big cities."
    bad2 = "Public transport is cheaper than owning a car."
    v2 = validate_candidate(src2, bad2, min_meaning=0.5)
    assert not v2.ok
    assert "key_content" in v2.reasons or "length" in v2.reasons
    print("cities drop reject OK:", v2.reasons)


def test_rejects_save_spend_flip() -> None:
    src = "Many students prefer to study from home because it saves time and money."
    bad = (
        "Many students prefer to study from home because it allows them to "
        "spend their time and money."
    )
    from app.pipeline.meaning_safety import polarity_safe

    assert not polarity_safe(src, bad)
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "polarity" in v.reasons
    print("save/spend polarity reject OK:", v.reasons)


def test_rejects_garbage_dontstet() -> None:
    src = (
        "If companies don't reduce packaging and people don't recycle properly, "
        "the problem will become worse every year."
    )
    bad = (
        "If companies don't reduce packaging, people don'tstet recycle properly, "
        "the problem will grow."
    )
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "broken" in v.reasons or "key_content" in v.reasons
    print("dontstet garbage reject OK:", v.reasons)


def test_rejects_recycle_negation_flip() -> None:
    src = (
        "If companies don't reduce packaging and people don't recycle properly, "
        "the problem will become worse every year."
    )
    bad = (
        "If people recycle and companies don't reduce their packaging and products, "
        "the problem will become worse every year."
    )
    from app.pipeline.meaning_safety import polarity_safe

    assert not polarity_safe(src, bad)
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "polarity" in v.reasons
    print("recycle negation flip reject OK:", v.reasons)


def test_assemble_preserves_two_units() -> None:
    from app.pipeline.orchestrator import _assemble_hybrid_units

    units = [
        (0, "Reading books regularly improves vocabulary and critical thinking."),
        (0, "Children who don't read enough often struggle with writing essays later in school."),
    ]
    outs = [
        "Reading books regularly improves vocabulary and critical thinking",
        "Children who don't read enough often struggle with writing essays later in school",
    ]
    text = _assemble_hybrid_units(units, outs)
    assert "books" in text.lower()
    assert "children" in text.lower()
    assert text.count(".") >= 2
    print("assemble two units OK:", text)


def test_rejects_comparison_swap() -> None:
    src = "Electric cars produce less air pollution than petrol cars."
    bad = "Petrol cars make less air pollution than electric cars."
    from app.pipeline.meaning_safety import polarity_safe

    assert not polarity_safe(src, bad)
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "polarity" in v.reasons
    ok, reasons = _accept_lm_unit(src, bad)
    assert not ok
    print("comparison swap reject OK:", v.reasons, reasons)


def test_rejects_learning_to_read_invention() -> None:
    src = "Learning a second language opens more career opportunities."
    bad = "Learning to read a second language opens many career opportunities."
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "invention" in v.reasons
    print("learning-to-read invention reject OK:", v.reasons)


def test_rejects_check_with_teachers_invention() -> None:
    src = (
        "Teachers should check progress regularly instead of grading only "
        "the final report."
    )
    bad = (
        "Teachers should check with teachers about progress instead of "
        "grading the final report."
    )
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "invention" in v.reasons
    print("check-with-teachers invention reject OK:", v.reasons)


def test_rejects_libraries_to_schools_swap() -> None:
    src = (
        "If cities don't fund them properly, many young people lose access "
        "to learning resources."
    )
    bad = (
        "If cities don't fund their schools correctly, many young people lose "
        "access to learning resources."
    )
    v = validate_candidate(src, bad, min_meaning=0.5)
    assert not v.ok
    assert "entity_inject" in v.reasons
    print("libraries/them->schools inject reject OK:", v.reasons)

    src2 = "Public libraries offer free books and quiet spaces for study."
    bad2 = "Public schools offer free books and quiet spaces for study."
    v2 = validate_candidate(src2, bad2, min_meaning=0.5)
    assert not v2.ok
    assert "topic_anchor" in v2.reasons or "key_content" in v2.reasons
    print("libraries->schools topic reject OK:", v2.reasons)


def test_clamp_rejects_multi_sentence() -> None:
    from app.pipeline.orchestrator import (
        _accept_lm_unit,
        _clamp_to_source_shape,
        _sentence_count_simple,
    )

    src = "Learning a second language opens more career opportunities."
    bad = (
        "Learning a second language opens more career opportunities. "
        "Travel becomes easier too."
    )
    clamped = _clamp_to_source_shape(src, bad)
    assert _sentence_count_simple(clamped) == 1
    ok, reasons = _accept_lm_unit(src, bad)
    assert not ok
    assert "shape" in reasons
    print("multi-sentence clamp/reject OK:", reasons)


if __name__ == "__main__":
    test_rejects_multi_sentence_invention()
    test_rejects_two_sentence_for_one()
    test_skip_short_and_light()
    test_sleep_style_must_visibly_rewrite()
    test_exercise_tv_must_not_be_reorder_only()
    test_complete_wording_rewrite_not_shuffle()
    test_restructure_time_only_and_manner()
    test_require_visible_keeps_structural_reorder()
    test_rejects_degree_adv_junk_reorder()
    test_because_front_preserves_clauses()
    test_esl_wording_avoids_synonym_thrash()
    test_rejects_hollow_ons_bills()
    test_assembly_preserves_unit_count()
    test_live_bug_patterns_rejected()
    test_rejects_fast_food_collocation_drop()
    test_rejects_workout_program_invention()
    test_keeps_fast_food_when_preserved()
    test_rejects_live_mangled_online_learning()
    test_rejects_dropped_online_modifier()
    test_rejects_mangled_climate_claim()
    test_accepts_faithful_paraphrase()
    test_rejects_dropped_books_and_cities()
    test_rejects_save_spend_flip()
    test_rejects_garbage_dontstet()
    test_rejects_recycle_negation_flip()
    test_assemble_preserves_two_units()
    test_rejects_comparison_swap()
    test_rejects_learning_to_read_invention()
    test_rejects_check_with_teachers_invention()
    test_rejects_libraries_to_schools_swap()
    test_clamp_rejects_multi_sentence()
    print("\nALL PHASE A TESTS PASSED")