tests/test_validation.py

from __future__ import annotations

from pathlib import Path

from gcmd_classifier.classification import (
    ClassificationCandidate,
    TermBranchSeed,
    VariableTerminalOutcome,
    candidate_from_term_branch,
    validate_candidate,
    validate_candidates,
    validate_terminal_outcomes,
)
from gcmd_classifier.models import ClassificationFinalStatus, SupportType
from gcmd_classifier.vocabulary import load_vocabulary

FIXTURE_PATH = Path("tests/fixtures/gcmd_hierarchy_small.json")
FULL_HIERARCHY_PATH = Path("data/gcmd_hierarchy.json")


def _index():
    return load_vocabulary(FIXTURE_PATH)


def _candidate(uuid: str, *, branch_id: str | None = "branch:one") -> ClassificationCandidate:
    index = _index()
    record = index.get(uuid)
    topic = record if record.level == "Topic" else index.get(index.ancestors_of(uuid)[-1])
    term = None
    if record.level == "Term":
        term = record
    elif record.level.startswith("Variable_Level_"):
        term = next(
            index.get(ancestor)
            for ancestor in index.ancestors_of(uuid)
            if index.get(ancestor).level == "Term"
        )
    return ClassificationCandidate(
        final_uuid=record.UUID,
        final_name=record.name,
        final_level=record.level,
        final_canonical_path=record.canonical_path,
        path_components=record.path_components,
        topic_uuid=topic.UUID,
        topic_name=topic.name,
        term_uuid=None if term is None else term.UUID,
        term_name=None if term is None else term.name,
        branch_id=branch_id,
        evidence="Validated evidence.",
        support_type=SupportType.EXPLICIT,
        confidence=0.81,
        reason="Routing selected the concept.",
        candidate_id="candidate_0001",
    )


def _term_branch() -> TermBranchSeed:
    return TermBranchSeed(
        branch_id="topic:topic_0001/term:term_0001",
        parent_topic_uuid="topic-atmosphere",
        parent_topic_name="ATMOSPHERE",
        term_uuid="term-atmospheric-chemistry",
        term_name="ATMOSPHERIC CHEMISTRY",
        term_level="Term",
        term_canonical_path="ATMOSPHERE > ATMOSPHERIC CHEMISTRY",
        evidence="The article supports the Term.",
        support_type=SupportType.EXPLICIT,
        confidence=0.77,
        reason="Term was selected.",
        candidate_id="term_0001",
        parent_branch_id="topic:topic_0001",
        prompt_version="term-test",
        model_provider="fake",
        model_name="fake-model",
        retry_count=0,
    )


def _terminal(
    uuid: str,
    *,
    branch_id: str = "topic:t/term:x/variable:y",
) -> VariableTerminalOutcome:
    index = _index()
    record = index.get(uuid)
    return VariableTerminalOutcome(
        branch_id=branch_id,
        parent_branch_id="topic:t/term:x",
        topic_uuid="topic-atmosphere",
        topic_name="ATMOSPHERE",
        term_uuid="term-atmospheric-chemistry",
        term_name="ATMOSPHERIC CHEMISTRY",
        final_uuid=record.UUID,
        final_name=record.name,
        final_level=record.level,
        final_canonical_path=record.canonical_path,
        path_components=record.path_components,
        evidence="Final-stage evidence.",
        support_type=SupportType.MIXED,
        confidence=0.69,
        reason="Selected descendant.",
        stop_reason="Deepest supported concept.",
        candidate_id="variable_0001",
        prompt_version="variable-test",
        model_provider="fake",
        model_name="fake-model",
        retry_count=1,
    )


def _error_codes(result) -> set[str]:
    return {error.code for error in result.deterministic_validation.errors}


def test_valid_candidate_accepted_and_populated_from_vocabulary_index() -> None:
    result = validate_candidate(_candidate("vl3-carbon-dioxide-profiles"), _index())

    assert result.valid is True
    assert result.classification is not None
    classification = result.classification
    assert classification.UUID == "vl3-carbon-dioxide-profiles"
    assert classification.name == "CARBON DIOXIDE PROFILES"
    assert classification.level == "Variable_Level_3"
    assert classification.canonical_path.endswith("CARBON DIOXIDE PROFILES")
    assert classification.topic == "ATMOSPHERE"
    assert classification.term == "ATMOSPHERIC CHEMISTRY"
    assert classification.parent_uuid == "vl2-carbon-dioxide"
    assert classification.final_status is ClassificationFinalStatus.ACCEPTED
    assert classification.deterministic_validation.valid is True
    assert classification.confidence is not None
    assert classification.confidence.variable_level_3 == 0.81
    assert classification.confidence.final == 0.81
    assert classification.classifier_evidence == "Validated evidence."
    assert classification.support_type is SupportType.EXPLICIT


def test_nonexistent_uuid_rejected() -> None:
    candidate = ClassificationCandidate(final_uuid="missing-uuid", final_level="Term")

    result = validate_candidate(candidate, _index())

    assert result.valid is False
    assert result.classification is None
    assert _error_codes(result) == {"UNKNOWN_UUID"}


def test_root_category_rejected() -> None:
    candidate = ClassificationCandidate(
        final_uuid="EARTH SCIENCE",
        final_name="EARTH SCIENCE",
        final_level="Category",
    )

    result = validate_candidate(candidate, _index())

    assert result.valid is False
    assert "ROOT_CATEGORY_NOT_ASSIGNABLE" in _error_codes(result)


def test_uuid_name_mismatch_rejected() -> None:
    candidate = _candidate("vl2-carbon-dioxide").model_copy(update={"final_name": "METHANE"})

    result = validate_candidate(candidate, _index())

    assert "UUID_NAME_MISMATCH" in _error_codes(result)
    assert result.classification is None


def test_uuid_level_mismatch_rejected() -> None:
    candidate = _candidate("vl2-carbon-dioxide").model_copy(
        update={"final_level": "Variable_Level_1"}
    )

    result = validate_candidate(candidate, _index())

    assert "UUID_LEVEL_MISMATCH" in _error_codes(result)


def test_uuid_canonical_path_mismatch_rejected() -> None:
    candidate = _candidate("vl2-carbon-dioxide").model_copy(
        update={"final_canonical_path": "ATMOSPHERE > MADE UP"}
    )

    result = validate_candidate(candidate, _index())

    assert "UUID_CANONICAL_PATH_MISMATCH" in _error_codes(result)


def test_invalid_path_components_rejected() -> None:
    candidate = _candidate("vl2-carbon-dioxide").model_copy(
        update={"path_components": ("ATMOSPHERE", "MADE UP")}
    )

    result = validate_candidate(candidate, _index())

    assert "PATH_COMPONENTS_MISMATCH" in _error_codes(result)


def test_invalid_parent_child_chain_rejected() -> None:
    candidate = _candidate("vl2-carbon-dioxide").model_copy(
        update={"expected_parent_uuid": "term-weather-events"}
    )

    result = validate_candidate(candidate, _index())

    assert "INVALID_PARENT_CHILD_CHAIN" in _error_codes(result)


def test_wrong_topic_membership_rejected() -> None:
    candidate = _candidate("vl2-carbon-dioxide").model_copy(
        update={"topic_uuid": "topic-oceans", "topic_name": "OCEANS"}
    )

    result = validate_candidate(candidate, _index())

    assert "WRONG_TOPIC_MEMBERSHIP" in _error_codes(result)
    assert "TOPIC_NAME_MISMATCH" in _error_codes(result)


def test_wrong_term_membership_rejected() -> None:
    candidate = _candidate("vl2-carbon-dioxide").model_copy(
        update={"term_uuid": "term-weather-events", "term_name": "WEATHER EVENTS"}
    )

    result = validate_candidate(candidate, _index())

    assert "WRONG_TERM_MEMBERSHIP" in _error_codes(result)
    assert "TERM_NAME_MISMATCH" in _error_codes(result)


def test_final_concept_under_expected_topic_and_term_succeeds() -> None:
    result = validate_candidate(_candidate("vl2-carbon-dioxide"), _index())

    assert result.valid is True
    assert result.classification is not None
    assert result.classification.UUID == "vl2-carbon-dioxide"


def test_topic_level_final_classification_succeeds() -> None:
    result = validate_candidate(_candidate("topic-atmosphere"), _index())

    assert result.valid is True
    assert result.classification is not None
    assert result.classification.level == "Topic"
    assert result.classification.term is None


def test_term_level_final_classification_succeeds_from_term_branch() -> None:
    result = validate_candidate(candidate_from_term_branch(_term_branch()), _index())

    assert result.valid is True
    assert result.classification is not None
    assert result.classification.level == "Term"
    assert result.classification.UUID == "term-atmospheric-chemistry"


def test_variable_level_1_final_classification_succeeds() -> None:
    assert validate_candidate(_candidate("vl1-atmosphere-carbon"), _index()).valid is True


def test_variable_level_2_final_classification_succeeds() -> None:
    assert validate_candidate(_candidate("vl2-carbon-dioxide"), _index()).valid is True


def test_variable_level_3_final_classification_succeeds() -> None:
    assert validate_candidate(_candidate("vl3-carbon-dioxide-profiles"), _index()).valid is True


def test_rejected_invalid_candidates_do_not_appear_in_accepted_classifications() -> None:
    batch = validate_candidates(
        [
            _candidate("vl2-carbon-dioxide"),
            ClassificationCandidate(final_uuid="missing-uuid", final_level="Term"),
        ],
        _index(),
    )

    assert [record.UUID for record in batch.accepted] == ["vl2-carbon-dioxide"]
    assert batch.rejected_count == 1
    assert batch.rejected[0].deterministic_validation.errors[0].details["field"] == "final_uuid"


def test_stop_at_term_terminal_outcome_can_become_valid_term_classification() -> None:
    terminal = _terminal("term-atmospheric-chemistry", branch_id="topic:t/term:x")

    result = validate_terminal_outcomes([terminal], _index())

    assert result.accepted_count == 1
    assert result.accepted[0].level == "Term"


def test_stop_at_variable_level_1_terminal_outcome_can_become_valid_classification() -> None:
    result = validate_terminal_outcomes([_terminal("vl1-atmosphere-carbon")], _index())

    assert result.accepted[0].level == "Variable_Level_1"


def test_selected_variable_level_3_terminal_outcome_can_become_valid_classification() -> None:
    result = validate_terminal_outcomes([_terminal("vl3-carbon-dioxide-profiles")], _index())

    assert result.accepted[0].level == "Variable_Level_3"
    assert result.accepted[0].reason_for_stopping == "Deepest supported concept."


def test_multiple_terminal_outcomes_validate_independently() -> None:
    result = validate_terminal_outcomes(
        [
            _terminal("vl2-carbon-dioxide", branch_id="topic:t/term:x/variable:a"),
            _terminal("vl2-methane", branch_id="topic:t/term:x/variable:b"),
        ],
        _index(),
    )

    assert [record.UUID for record in result.accepted] == ["vl2-carbon-dioxide", "vl2-methane"]


def test_validation_does_not_require_model_call() -> None:
    result = validate_candidate(_candidate("vl2-methane"), _index())

    assert result.valid is True


def test_full_data_representative_records_pass_validation() -> None:
    index = load_vocabulary(FULL_HIERARCHY_PATH)
    representatives = []
    for level in ["Topic", "Term", "Variable_Level_1", "Variable_Level_2", "Variable_Level_3"]:
        representatives.append(next(record for record in index.records if record.level == level))

    for record in representatives:
        topic = (
            record if record.level == "Topic" else index.get(index.ancestors_of(record.UUID)[-1])
        )
        term = None
        if record.level == "Term":
            term = record
        elif record.level.startswith("Variable_Level_"):
            term = next(
                index.get(ancestor)
                for ancestor in index.ancestors_of(record.UUID)
                if index.get(ancestor).level == "Term"
            )
        candidate = ClassificationCandidate(
            final_uuid=record.UUID,
            final_name=record.name,
            final_level=record.level,
            final_canonical_path=record.canonical_path,
            path_components=record.path_components,
            topic_uuid=topic.UUID,
            topic_name=topic.name,
            term_uuid=None if term is None else term.UUID,
            term_name=None if term is None else term.name,
        )
        assert validate_candidate(candidate, index).valid is True