"""Adversarial regression cases for `rag.grounding.verify`.

Each case reproduces a defect found by Codex's 2026-08-06 code review
(`coordination/CODEX_RAG_CODE_REVIEW_2026-08-06.md`, F-01) against the old
implementation, which pooled every evidence number into one global set and
never required a citation at all. `verify` must now bind a claim's numbers
to only the evidence block(s) its own citation group names, and must reject
a claim with no citation regardless of whether it contains a number.
"""
from rag import grounding


def test_number_from_the_wrong_evidence_block_is_rejected():
    # Reproduces `so_sai_nguon`: 500 mg is real, but only in evidence 2 —
    # citing [1] for it must fail, not pass because 500 exists *somewhere*.
    report = grounding.verify(
        "Liều 500 mg [1].",
        ("Không dùng khi suy thận.", "Liều 500 mg mỗi ngày."),
    )
    assert not report.grounded
    assert report.unsupported_numbers == ("500",)
    assert report.reason == "ungrounded_number"


def test_citing_the_correct_block_for_the_number_is_grounded():
    report = grounding.verify(
        "Liều 500 mg [2].",
        ("Không dùng khi suy thận.", "Liều 500 mg mỗi ngày."),
    )
    assert report.grounded
    assert report.cited_indices == (2,)


def test_answer_with_no_citation_at_all_is_rejected():
    # Reproduces `khong_citation`: a number that is genuinely in the evidence
    # still must not pass when the answer never cites anything.
    report = grounding.verify("Liều 500 mg.", ("Liều 500 mg mỗi ngày.",))
    assert not report.grounded
    assert report.uncited_claim
    assert report.unsupported_numbers == ("500",)
    assert report.reason == "ungrounded_number"


def test_nonnumeric_claim_with_no_citation_is_rejected():
    report = grounding.verify(
        "Chống chỉ định với suy gan nặng.", ("Chống chỉ định: suy gan nặng.",)
    )
    assert not report.grounded
    assert report.uncited_claim
    assert report.reason == "uncited_claim"


def test_trailing_text_after_the_last_citation_needs_its_own_citation():
    report = grounding.verify(
        "Liều 500 mg [1]. Không dùng khi suy thận.",
        ("Liều 500 mg mỗi ngày.",),
    )
    assert not report.grounded
    assert report.uncited_claim


def test_out_of_range_citation_is_invalid_and_leaves_its_claim_unsupported():
    report = grounding.verify("Liều 500 mg [3].", ("Liều 500 mg mỗi ngày.",))
    assert not report.grounded
    assert report.invalid_citations == (3,)
    assert report.unsupported_numbers == ("500",)
    assert report.reason == "ungrounded_number"


def test_two_claims_each_binding_correctly_to_their_own_source_is_grounded():
    report = grounding.verify(
        "Người lớn 500 mg [1]. Trẻ em 250 mg [2].",
        ("Liều người lớn 500 mg.", "Liều trẻ em 250 mg."),
    )
    assert report.grounded
    assert report.cited_indices == (1, 2)


def test_second_claim_citing_the_first_blocks_source_number_is_rejected():
    # The child dose (250) is real, but only in evidence 2; citing [1] for it
    # is the same defect as `so_sai_nguon`, just in a second sentence.
    report = grounding.verify(
        "Người lớn 500 mg [1]. Trẻ em 250 mg [1].",
        ("Liều người lớn 500 mg.", "Liều trẻ em 250 mg."),
    )
    assert not report.grounded
    assert report.unsupported_numbers == ("250",)


def test_multiple_markers_on_one_claim_check_against_their_union():
    report = grounding.verify(
        "Liều 500 mg [1][2].",
        ("Liều người lớn 500 mg.", "Liều mỗi ngày."),
    )
    assert report.grounded
    assert report.cited_indices == (1, 2)


def test_extractive_quoting_format_is_still_grounded():
    # The service's extractive fallback formats each block as `text [n]`.
    report = grounding.verify(
        "Liều được ghi trong nguồn. [1]", ("Liều được ghi trong nguồn.",)
    )
    assert report.grounded


def test_decimal_separator_is_compared_verbatim_not_normalised():
    report = grounding.verify("Liều 7.5 mg [1].", ("Liều 7,5 mg.",))
    assert not report.grounded
    assert report.unsupported_numbers == ("7.5",)


def test_known_gap_fabricated_nonnumeric_claim_with_a_valid_citation_still_passes():
    # Documents the residual gap `grounding.verify` cannot close on its own
    # (see module docstring): a citation-bearing claim whose content the
    # cited block does not actually support. Closed by a separate LLM
    # entailment pass in `rag/answer.py`, not by this regex-only check.
    report = grounding.verify(
        "Metformin chữa ung thư [1].",
        ("Metformin dùng điều trị đái tháo đường.",),
    )
    assert report.grounded