Metamorph/backend/tests/test_schema.py

"""M1 schema integration test.

Asserts that the migration has produced the expected tables, FK relations,
CHECK constraints, partial indexes, and that the alembic_version row is at head.
Skips automatically when Postgres is unreachable.
"""

from __future__ import annotations

import pytest
from sqlalchemy import inspect, text

EXPECTED_TABLES = {
    # Auth / RBAC
    "users",
    "groups",
    "permissions",
    "user_groups",
    "group_permissions",
    "invitations",
    "invitation_groups",
    "refresh_tokens",
    # MITRE
    "mitre_tactics",
    "mitre_techniques",
    "mitre_subtechniques",
    "mitre_technique_tactics",
    # Templates
    "test_templates",
    "test_template_mitre_tags",
    "scenario_templates",
    "scenario_template_tests",
    # Missions
    "missions",
    "mission_members",
    "mission_scenarios",
    "mission_tests",
    "mission_test_mitre_tags",
    "mission_categories",
    # Evidence / settings / notifications
    "evidence_files",
    "settings",
    "detection_levels",
    "notifications",
    # Alembic bookkeeping
    "alembic_version",
}

# Tables that MUST carry a `deleted_at` column (soft delete).
SOFT_DELETE_TABLES = {
    "users",
    "groups",
    "test_templates",
    "scenario_templates",
    "missions",
    "mission_scenarios",
    "mission_tests",
    "mission_categories",
    "evidence_files",
}

# Tables that MUST carry the standard `created_at` + `updated_at` pair.
TIMESTAMP_TABLES = {
    "users",
    "groups",
    "test_templates",
    "scenario_templates",
    "missions",
    "mission_scenarios",
    "mission_tests",
    "mission_categories",
    "evidence_files",
}

# Spot-checked FK pairs (child_table, child_col, parent_table).
EXPECTED_FKS = {
    ("evidence_files", "mission_test_id", "mission_tests"),
    ("evidence_files", "uploaded_by_user_id", "users"),
    ("mission_members", "mission_id", "missions"),
    ("mission_members", "user_id", "users"),
    ("mission_scenarios", "mission_id", "missions"),
    ("mission_tests", "scenario_id", "mission_scenarios"),
    ("mission_tests", "detection_level_id", "detection_levels"),
    ("group_permissions", "group_id", "groups"),
    ("group_permissions", "permission_id", "permissions"),
    ("user_groups", "user_id", "users"),
    ("user_groups", "group_id", "groups"),
    ("refresh_tokens", "user_id", "users"),
    ("notifications", "user_id", "users"),
    ("mitre_subtechniques", "technique_id", "mitre_techniques"),
}

# CHECK constraint names we expect to see (namespace 'public' only).
# `mission_test_mitre_tags` deliberately lacks the exactly_one_mitre_fk check
# because it is denormalised — see app/models/mission.py docstring.
EXPECTED_CHECKS = {
    "ck_missions_status_valid",
    "ck_missions_visibility_mode_valid",
    "ck_mission_tests_state_valid",
    "ck_mission_tests_snapshot_opsec_level_valid",
    "ck_test_templates_opsec_level_valid",
    "ck_mission_members_role_hint_valid",
    "ck_test_template_mitre_tags_mitre_kind_valid",
    "ck_test_template_mitre_tags_exactly_one_mitre_fk",
    "ck_mission_test_mitre_tags_mitre_kind_valid",
}


@pytest.fixture(scope="module")
def insp(db_engine_or_skip):
    return inspect(db_engine_or_skip)


def test_all_expected_tables_exist(insp):
    actual = set(insp.get_table_names(schema="public"))
    missing = EXPECTED_TABLES - actual
    assert not missing, f"missing tables: {sorted(missing)}"


def test_soft_delete_columns_present(insp):
    for tbl in sorted(SOFT_DELETE_TABLES):
        cols = {c["name"] for c in insp.get_columns(tbl)}
        assert "deleted_at" in cols, f"{tbl} missing deleted_at"


def test_standard_timestamp_columns_present(insp):
    for tbl in sorted(TIMESTAMP_TABLES):
        cols = {c["name"] for c in insp.get_columns(tbl)}
        assert "created_at" in cols, f"{tbl} missing created_at"
        assert "updated_at" in cols, f"{tbl} missing updated_at"


def test_partial_index_for_soft_delete(db_engine_or_skip):
    """Each soft-delete table must carry an `ix_<table>_active` partial index."""
    with db_engine_or_skip.connect() as conn:
        rows = conn.execute(
            text(
                "SELECT indexname FROM pg_indexes "
                "WHERE schemaname='public' AND indexdef ILIKE '%deleted_at IS NULL%'"
            )
        ).all()
    names = {r[0] for r in rows}
    for tbl in SOFT_DELETE_TABLES:
        assert f"ix_{tbl}_active" in names, f"{tbl}: partial index missing — got {names}"


def test_expected_foreign_keys(insp):
    all_fks = set()
    for tbl in EXPECTED_TABLES:
        if tbl == "alembic_version":
            continue
        for fk in insp.get_foreign_keys(tbl):
            for col in fk["constrained_columns"]:
                all_fks.add((tbl, col, fk["referred_table"]))
    for triple in EXPECTED_FKS:
        assert triple in all_fks, f"missing FK: {triple}"


def test_expected_check_constraints(db_engine_or_skip):
    with db_engine_or_skip.connect() as conn:
        rows = conn.execute(
            text(
                "SELECT conname FROM pg_constraint "
                "WHERE contype='c' AND connamespace = 'public'::regnamespace"
            )
        ).all()
    names = {r[0] for r in rows}
    missing = EXPECTED_CHECKS - names
    assert not missing, f"missing CHECK constraints: {sorted(missing)}"


def test_alembic_at_head(db_engine_or_skip):
    """The DB must be at the latest migration after `make migrate`."""
    with db_engine_or_skip.connect() as conn:
        rev = conn.execute(text("SELECT version_num FROM alembic_version")).scalar_one()
    assert rev, "alembic_version is empty — migrate didn't run"
    assert len(rev) >= 8, f"unexpected alembic version: {rev}"


def test_exactly_one_mitre_fk_check_enforced(db_engine_or_skip):
    """Inserting a tag with two non-null FKs must raise (CHECK constraint)."""
    import uuid

    from sqlalchemy.exc import IntegrityError

    with db_engine_or_skip.begin() as conn:
        # Seed a minimal test_template + tactic + technique to reference.
        tmpl_id = uuid.uuid4()
        tactic_id = uuid.uuid4()
        technique_id = uuid.uuid4()
        conn.execute(
            text(
                "INSERT INTO test_templates (id, name, opsec_level) "
                "VALUES (:id, 'tmp', 'low')"
            ),
            {"id": tmpl_id},
        )
        conn.execute(
            text(
                "INSERT INTO mitre_tactics (id, external_id, short_name, name) "
                "VALUES (:id, 'TA0099', 'tmp', 'tmp')"
            ),
            {"id": tactic_id},
        )
        conn.execute(
            text(
                "INSERT INTO mitre_techniques (id, external_id, name) "
                "VALUES (:id, 'T9999', 'tmp')"
            ),
            {"id": technique_id},
        )

    # Now try to insert a violating row — must fail.
    with pytest.raises(IntegrityError):
        with db_engine_or_skip.begin() as conn:
            conn.execute(
                text(
                    "INSERT INTO test_template_mitre_tags "
                    "(id, test_template_id, mitre_kind, tactic_id, technique_id) "
                    "VALUES (:id, :tmpl, 'tactic', :tac, :tech)"
                ),
                {
                    "id": uuid.uuid4(),
                    "tmpl": tmpl_id,
                    "tac": tactic_id,
                    "tech": technique_id,
                },
            )

    # Cleanup so the test is rerunnable.
    with db_engine_or_skip.begin() as conn:
        conn.execute(text("DELETE FROM mitre_techniques WHERE id = :id"), {"id": technique_id})
        conn.execute(text("DELETE FROM mitre_tactics WHERE id = :id"), {"id": tactic_id})
        conn.execute(text("DELETE FROM test_templates WHERE id = :id"), {"id": tmpl_id})