test_model.py

# ruff: noqa: F811
import json
import typing as t
from datetime import date, datetime
from pathlib import Path
from unittest.mock import patch, PropertyMock

import time_machine
import pandas as pd  # noqa: TID253
import pytest
from pytest_mock.plugin import MockerFixture
from sqlglot import exp, parse_one
from sqlglot.errors import ParseError
from sqlglot.schema import MappingSchema
from sqlmesh.cli.project_init import init_example_project, ProjectTemplate
from sqlmesh.core.environment import EnvironmentNamingInfo
from sqlmesh.core.model.kind import TimeColumn, ModelKindName

from sqlmesh import CustomMaterialization, CustomKind
from pydantic import model_validator, ValidationError
from sqlmesh.core import constants as c
from sqlmesh.core import dialect as d
from sqlmesh.core.console import get_console
from sqlmesh.core.audit import ModelAudit, load_audit
from sqlmesh.core.model.common import ParsableSql
from sqlmesh.core.config import (
    Config,
    DuckDBConnectionConfig,
    GatewayConfig,
    NameInferenceConfig,
    ModelDefaultsConfig,
    LinterConfig,
)
from sqlmesh.core import constants as c
from sqlmesh.core.context import Context, ExecutionContext
from sqlmesh.core.dialect import parse
from sqlmesh.core.engine_adapter.base import MERGE_SOURCE_ALIAS, MERGE_TARGET_ALIAS
from sqlmesh.core.engine_adapter.duckdb import DuckDBEngineAdapter
from sqlmesh.core.macros import MacroEvaluator, macro
from sqlmesh.core.model import (
    CustomKind,
    PythonModel,
    FullKind,
    IncrementalByTimeRangeKind,
    IncrementalUnmanagedKind,
    IncrementalByUniqueKeyKind,
    ModelCache,
    ModelMeta,
    SeedKind,
    SqlModel,
    TimeColumn,
    ExternalKind,
    ViewKind,
    create_external_model,
    create_seed_model,
    create_sql_model,
    load_sql_based_model,
    load_sql_based_models,
    model,
)
from sqlmesh.core.model.common import parse_expression
from sqlmesh.core.model.kind import ModelKindName, _model_kind_validator
from sqlmesh.core.model.seed import CsvSettings
from sqlmesh.core.node import IntervalUnit, _Node
from sqlmesh.core.signal import signal
from sqlmesh.core.snapshot import Snapshot, SnapshotChangeCategory
from sqlmesh.utils.date import TimeLike, to_datetime, to_ds, to_timestamp
from sqlmesh.utils.errors import ConfigError, SQLMeshError, LinterError
from sqlmesh.utils.jinja import JinjaMacroRegistry, MacroInfo, MacroExtractor
from sqlmesh.utils.metaprogramming import Executable, SqlValue
from sqlmesh.core.macros import RuntimeStage
from tests.utils.test_helpers import use_terminal_console


def missing_schema_warning_msg(model, deps):
    deps = ", ".join(f"'{dep}'" for dep in sorted(deps))
    return (
        f"SELECT * cannot be expanded due to missing schema(s) for model(s): {deps}. "
        f"Run `sqlmesh create_external_models` and / or make sure that the model '{model}' "
        "can be rendered at parse time."
    )


def test_load(assert_exp_eq):
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name,
            storage_format iceberg,
            partitioned_by d,
            clustered_by e,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column a,
            ),
            tags [tag_foo, tag_bar],
            grain [a, b],
            references (
                f,
                g
            ),
        );

        @DEF(x, 1);
        @DEF(y, @x + 1);
        CACHE TABLE x AS SELECT 1;
        ADD JAR 's3://my_jar.jar';

        SELECT
            1::int AS a,
            CAST(2 AS double) AS b,
            c::bool,
            1::int AS d, -- d
            CAST(2 AS double) AS e, --e
            f::bool, --f
            @y::int AS g,
        FROM
            db.other_table t1
            LEFT JOIN
            db.table t2
            ON
                t1.a = t2.a;

        DROP TABLE x;
    """
    )

    model = load_sql_based_model(expressions)
    assert model.name == "db.table"
    assert model.owner == "owner_name"
    assert model.dialect == "spark"
    assert model.table_format is None
    assert model.storage_format == "iceberg"
    assert [col.sql() for col in model.partitioned_by] == ['"a"', '"d"']
    assert [col.sql() for col in model.clustered_by] == ['"e"']
    assert model.columns_to_types == {
        "a": exp.DataType.build("int"),
        "b": exp.DataType.build("double"),
        "c": exp.DataType.build("boolean"),
        "d": exp.DataType.build("int"),
        "e": exp.DataType.build("double"),
        "f": exp.DataType.build("boolean"),
        "g": exp.DataType.build("int"),
    }
    assert model.annotated
    assert model.view_name == "table"
    assert model.macro_definitions == [d.parse_one("@DEF(x, 1)"), d.parse_one("@DEF(y, @x + 1)")]
    assert list(model.pre_statements) == [
        d.parse_one("@DEF(x, 1)"),
        d.parse_one("@DEF(y, @x + 1)"),
        d.parse_one("CACHE TABLE x AS SELECT 1"),
        d.parse_one("ADD JAR 's3://my_jar.jar'", dialect="spark"),
    ]
    assert list(model.post_statements) == [
        d.parse_one("DROP TABLE x"),
    ]
    assert model.depends_on == {'"db"."other_table"'}

    assert (
        model.render_query().sql(pretty=True, dialect="spark")
        == """SELECT
  CAST(1 AS INT) AS `a`,
  CAST(2 AS DOUBLE) AS `b`,
  CAST(`c` AS BOOLEAN) AS `c`,
  CAST(1 AS INT) AS `d`, /* d */
  CAST(2 AS DOUBLE) AS `e`, /* e */
  CAST(`f` AS BOOLEAN) AS `f`, /* f */
  CAST(1 + 1 AS INT) AS `g`
FROM `db`.`other_table` AS `t1`
LEFT JOIN `db`.`table` AS `t2`
  ON `t1`.`a` = `t2`.`a`"""
    )

    assert model.tags == ["tag_foo", "tag_bar"]
    assert [r.dict() for r in model.all_references] == [
        {"model_name": "db.table", "expression": d.parse_one("[a, b]"), "unique": True},
        {"model_name": "db.table", "expression": d.parse_one("f"), "unique": True},
        {"model_name": "db.table", "expression": d.parse_one("g"), "unique": True},
    ]


def test_model_multiple_select_statements():
    # Make sure the load_model raises an exception for model with multiple select statements.
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name,
        );

        SELECT 1, ds;
        SELECT 2, ds;
        """
    )
    with pytest.raises(ConfigError, match=r"^Only one SELECT.*"):
        load_sql_based_model(expressions)


def test_model_validation(tmp_path):
    expressions = d.parse(
        f"""
        MODEL (
            name db.table,
            kind FULL,
        );

        SELECT
          y::int,
          x::int AS y
        FROM db.ext
        """
    )

    ctx = Context(
        config=Config(linter=LinterConfig(enabled=True, rules=["noambiguousprojections"])),
        paths=tmp_path,
    )
    ctx.upsert_model(load_sql_based_model(expressions, default_catalog="memory"))

    errors = ctx.lint_models(["db.table"], raise_on_error=False)
    assert errors, "Expected NoAmbiguousProjections violation"
    assert errors[0].violation_msg == "Found duplicate outer select name 'y'"

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind FULL,
        );

        SELECT a, a UNION SELECT c, c
        """
    )

    ctx.upsert_model(load_sql_based_model(expressions, default_catalog="memory"))

    errors = ctx.lint_models(["db.table"], raise_on_error=False)
    assert errors, "Expected NoAmbiguousProjections violation"
    assert errors[0].violation_msg == "Found duplicate outer select name 'a'"

    expressions = d.parse(
        f"""
        MODEL (
            name db.table,
            kind FULL,
        );

        SELECT * FROM db.table
        """
    )

    model = load_sql_based_model(expressions)
    with pytest.raises(ConfigError) as ex:
        model.validate_definition()

    assert "require inferrable column types" in str(ex.value)


def test_model_union_query(sushi_context, assert_exp_eq):
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind FULL,
        );

        SELECT a, b UNION SELECT c, c
        """
    )

    load_sql_based_model(expressions)

    expressions = d.parse(
        """
        MODEL (
            name sushi.test,
            kind FULL,
        );

        @union('all', sushi.marketing, sushi.marketing)
        """
    )
    sushi_context.upsert_model(load_sql_based_model(expressions, default_catalog="memory"))
    assert_exp_eq(
        sushi_context.get_model("sushi.test").render_query(),
        """SELECT
  CAST("marketing"."customer_id" AS INT) AS "customer_id",
  CAST("marketing"."status" AS TEXT) AS "status",
  CAST("marketing"."updated_at" AS TIMESTAMPNTZ) AS "updated_at",
  CAST("marketing"."valid_from" AS TIMESTAMP) AS "valid_from",
  CAST("marketing"."valid_to" AS TIMESTAMP) AS "valid_to"
FROM "memory"."sushi"."marketing" AS "marketing"
UNION ALL
SELECT
  CAST("marketing"."customer_id" AS INT) AS "customer_id",
  CAST("marketing"."status" AS TEXT) AS "status",
  CAST("marketing"."updated_at" AS TIMESTAMPNTZ) AS "updated_at",
  CAST("marketing"."valid_from" AS TIMESTAMP) AS "valid_from",
  CAST("marketing"."valid_to" AS TIMESTAMP) AS "valid_to"
FROM "memory"."sushi"."marketing" AS "marketing"
        """,
    )


@time_machine.travel("1996-02-10 00:00:00 UTC")
@pytest.mark.parametrize(
    "test_id, condition, union_type, table_count, expected_result",
    [
        # Test case 1: Basic conditional union - True condition
        (
            "test_1",
            "@get_date() == '1996-02-10'",
            "'all'",
            2,
            lambda expected_select: f"{expected_select}\nUNION ALL\n{expected_select}\n",
        ),
        # Test case 2: False condition - should return just first table
        (
            "test_2",
            "@get_date() > '1996-02-10'",
            "'all'",
            2,
            lambda expected_select: f"{expected_select}\n",
        ),
        # Test case 3: Multiple tables in union
        (
            "test_3",
            "@get_date() == '1996-02-10'",
            "'all'",
            3,
            lambda expected_select: f"{expected_select}\nUNION ALL\n{expected_select}\nUNION ALL\n{expected_select}\n",
        ),
        # Test case 4: DISTINCT type
        (
            "test_4",
            "@get_date() == '1996-02-10'",
            "'distinct'",
            2,
            lambda expected_select: f"{expected_select}\nUNION\n{expected_select}\n",
        ),
        # Test case 5: Complex condition
        (
            "test_5",
            "@get_date() = '1996-02-10' and 1=1 or @get_date() > '1996-02-10'",
            "'distinct'",
            2,
            lambda expected_select: f"{expected_select}\nUNION\n{expected_select}\n",
        ),
        # Test case 6: Missing union type (defaults to ALL)
        (
            "test_6",
            "@get_date() == '1996-02-10'",
            "",
            2,
            lambda expected_select: f"{expected_select}\nUNION ALL\n{expected_select}\n",
        ),
        # Test case 7: Missing union type AND condition
        (
            "test_7",
            "",
            "",
            2,
            lambda expected_select: f"{expected_select}\nUNION ALL\n{expected_select}\n",
        ),
        # Test case 8: Missing union type AND condition multiple tables
        (
            "test_8",
            "",
            "",
            3,
            lambda expected_select: f"{expected_select}\nUNION ALL\n{expected_select}\n\nUNION ALL\n{expected_select}\n",
        ),
        # Test case 9: Missing union type AND condition one table
        (
            "test_9",
            "",
            "",
            1,
            lambda expected_select: f"{expected_select}",
        ),
        # Test case 10: Union type with one table
        (
            "test_10",
            "",
            "'distinct'",
            1,
            lambda expected_select: f"{expected_select}",
        ),
        # Test case 11: Condition with one table
        (
            "test_9",
            "True",
            "",
            1,
            lambda expected_select: f"{expected_select}",
        ),
    ],
)
def test_model_union_conditional(
    sushi_context, assert_exp_eq, test_id, condition, union_type, table_count, expected_result
):
    @macro()
    def get_date(evaluator):
        from sqlmesh.utils.date import now

        return f"'{now().date()}'"

    expected_select = """SELECT
  CAST("marketing"."customer_id" AS INT) AS "customer_id",
  CAST("marketing"."status" AS TEXT) AS "status",
  CAST("marketing"."updated_at" AS TIMESTAMPNTZ) AS "updated_at",
  CAST("marketing"."valid_from" AS TIMESTAMP) AS "valid_from",
  CAST("marketing"."valid_to" AS TIMESTAMP) AS "valid_to"
FROM "memory"."sushi"."marketing" AS "marketing"
"""

    # Create tables argument list based on table_count
    tables = ", ".join(["sushi.marketing"] * table_count)

    # Handle the missing union_type case
    union_type_arg = f", {union_type}" if union_type else ""

    expressions = d.parse(
        f"""
        MODEL (
            name sushi.{test_id},
            kind FULL,
        );

        @union({condition}{union_type_arg}, {tables})
        """
    )
    sushi_context.upsert_model(load_sql_based_model(expressions, default_catalog="memory"))

    assert_exp_eq(
        sushi_context.get_model(f"sushi.{test_id}").render_query(),
        expected_result(expected_select),
    )


@use_terminal_console
def test_model_qualification(tmp_path: Path):
    with patch.object(get_console(), "log_warning") as mock_logger:
        expressions = d.parse(
            """
            MODEL (
                name db.table,
                kind FULL,
            );

            SELECT a
            """
        )

        ctx = Context(
            config=Config(linter=LinterConfig(enabled=True, warn_rules=["ALL"])), paths=tmp_path
        )
        ctx.upsert_model(load_sql_based_model(expressions))
        ctx.plan_builder("dev")

        assert (
            """Column '"a"' could not be resolved for model '"db"."table"', the column may not exist or is ambiguous."""
            in mock_logger.call_args[0][0]
        )


@use_terminal_console
def test_model_missing_audits(tmp_path: Path):
    with patch.object(get_console(), "log_warning") as mock_logger:
        expressions = d.parse(
            """
            MODEL (
                name db.table,
                kind FULL,
            );

            SELECT a
            """
        )

        ctx = Context(
            config=Config(linter=LinterConfig(enabled=True, warn_rules=["nomissingaudits"])),
            paths=tmp_path,
        )
        ctx.upsert_model(load_sql_based_model(expressions))
        ctx.plan_builder("plan")

        assert (
            """Model `audits` must be configured to test data quality."""
            in mock_logger.call_args[0][0]
        )


def test_project_is_set_in_standalone_audit(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    db_path = str(tmp_path / "db.db")
    db_connection = DuckDBConnectionConfig(database=db_path)

    config = Config(
        project="test",
        gateways={"gw": GatewayConfig(connection=db_connection)},
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
    )

    model = tmp_path / "models" / "some_model.sql"
    model.parent.mkdir(parents=True, exist_ok=True)
    model.write_text("MODEL (name m); SELECT 1 AS c")

    audit = tmp_path / "audits" / "a_standalone_audit.sql"
    audit.parent.mkdir(parents=True, exist_ok=True)
    audit.write_text("AUDIT (name a, standalone true); SELECT * FROM m WHERE c <= 0")

    context = Context(paths=tmp_path, config=config)
    context.plan(no_prompts=True, auto_apply=True)

    model = tmp_path / "models" / "some_model.sql"
    model.parent.mkdir(parents=True, exist_ok=True)
    model.write_text("MODEL (name m); SELECT 2 AS c")

    assert context.fetchdf(
        "select snapshot -> 'node' -> 'project' AS standalone_audit_project "
        """from sqlmesh._snapshots where (snapshot -> 'node' -> 'source_type')::text = '"audit"'"""
    ).to_dict()["standalone_audit_project"] == {0: '"test"'}
    assert context.load().standalone_audits["a"].project == "test"


@pytest.mark.parametrize(
    "partition_by_input, partition_by_output, output_dialect, expected_exception",
    [
        ("a", ["`a`"], "bigquery", None),
        ("(a, b)", ["`a`", "`b`"], "bigquery", None),
        ("TIMESTAMP_TRUNC(`a`, DAY)", ["TIMESTAMP_TRUNC(`a`, DAY)"], "bigquery", None),
        ("e", "", "bigquery", ConfigError),
        ("(a, e)", "", "bigquery", ConfigError),
        ("(a, a)", "", "bigquery", ConfigError),
        ("(day(a),b)", ['DAY("a")', '"b"'], "trino", None),
    ],
)
def test_partitioned_by(
    partition_by_input, partition_by_output, output_dialect, expected_exception
):
    expressions = d.parse(
        f"""
        MODEL (
            name db.table,
            dialect bigquery,
            owner owner_name,
            partitioned_by {partition_by_input},
            clustered_by (c, d),
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column a,
            ),
        );

        SELECT 1::int AS a, 2::int AS b, 3 AS c, 4 as d;
    """
    )

    model = load_sql_based_model(expressions)
    assert model.clustered_by == [exp.to_column('"c"'), exp.to_column('"d"')]
    if expected_exception:
        with pytest.raises(expected_exception):
            model.validate_definition()
    else:
        model.validate_definition()
        assert [
            col.sql(dialect=output_dialect) for col in model.partitioned_by
        ] == partition_by_output


def test_opt_out_of_time_column_in_partitioned_by():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect bigquery,
            partitioned_by b,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column a,
                partition_by_time_column false
            ),
        );

        SELECT 1::int AS a, 2::int AS b;
    """
    )

    model = load_sql_based_model(expressions)
    assert model.partitioned_by == [exp.to_column('"b"')]


def test_model_no_name():
    expressions = d.parse(
        """
        MODEL (
            dialect bigquery,
        );

        SELECT 1::int AS a, 2::int AS b;
    """
    )

    with pytest.raises(ConfigError) as ex:
        load_sql_based_model(expressions)
    assert (
        str(ex.value)
        == "Please add the required 'name' field to the MODEL block at the top of the file.\n\nLearn more at https://sqlmesh.readthedocs.io/en/stable/concepts/models/overview"
    )


def test_model_field_name_suggestions():
    # top-level field
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialects bigquery,
        );

        SELECT 1::int AS a, 2::int AS b;
    """
    )

    with pytest.raises(ConfigError) as ex:
        load_sql_based_model(expressions)
    assert (
        str(ex.value)
        == "Invalid field name present in the MODEL block: 'dialects'. Did you mean 'dialect'?"
    )

    # kind field
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column a,
                batch_sizes 1
            ),
        );

        SELECT 1::int AS a, 2::int AS b;
    """
    )

    with pytest.raises(ConfigError) as ex:
        load_sql_based_model(expressions)
    assert (
        str(ex.value)
        == "Invalid field name present in the MODEL block 'kind INCREMENTAL_BY_TIME_RANGE' field: 'batch_sizes'. Did you mean 'batch_size'?"
    )

    # multiple fields
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialects bigquery,
            descriptions 'a',
            asdfasdf true
        );

        SELECT 1::int AS a, 2::int AS b;
    """
    )

    with pytest.raises(ConfigError) as ex:
        load_sql_based_model(expressions)
    ex_str = str(ex.value)
    # field order is non-deterministic, so we can't test the output string directly
    assert "Invalid field names present in the MODEL block: " in ex_str
    assert "'descriptions'" in ex_str
    assert "'dialects'" in ex_str
    assert "'asdfasdf'" in ex_str
    assert "- descriptions: Did you mean 'description'?" in ex_str
    assert "- dialects: Did you mean 'dialect'?" in ex_str
    assert "- asdfasdf: Did you mean " not in ex_str


def test_model_required_field_missing():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE (),
        );

        SELECT 1::int AS a, 2::int AS b;
    """
    )

    with pytest.raises(ConfigError) as ex:
        load_sql_based_model(expressions)
    assert (
        str(ex.value)
        == "Please add required field 'time_column' to the MODEL block 'kind INCREMENTAL_BY_TIME_RANGE' field."
    )


def test_no_model_statement(tmp_path: Path):
    # No name inference => MODEL (...) is required
    expressions = d.parse("SELECT 1 AS x")
    with pytest.raises(
        ConfigError,
        match="Please add a MODEL block at the top of the file. Example:",
    ):
        load_sql_based_model(expressions)

    # Name inference is enabled => MODEL (...) not required
    init_example_project(tmp_path, engine_type="duckdb")

    test_sql_file = tmp_path / "models/test_schema/test_model.sql"
    test_sql_file.parent.mkdir(parents=True, exist_ok=True)
    test_sql_file.write_text("SELECT 1 AS c")

    config = Config(
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
        model_naming=NameInferenceConfig(infer_names=True),
    )
    context = Context(paths=tmp_path, config=config)

    model = context.get_model("test_schema.test_model")
    assert isinstance(model, SqlModel)
    assert model.name == "test_schema.test_model"


def test_unordered_model_statements():
    expressions = d.parse(
        """
        SELECT 1 AS x;

        MODEL (
            name db.table,
            dialect spark,
            owner owner_name
        );
    """
    )

    with pytest.raises(ConfigError) as ex:
        load_sql_based_model(expressions)
    assert "Please add a MODEL block at the top of the file. Example:" in str(ex.value)


def test_no_query():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name
        );

        @DEF(x, 1)
        """
    )

    with pytest.raises(ConfigError) as ex:
        model = load_sql_based_model(expressions, path=Path("test_location"))
        model.validate_definition()

    assert "Model query needs to be a SELECT or a UNION, got @DEF(x, 1)." in str(ex.value)


def test_single_macro_as_query(assert_exp_eq):
    @macro()
    def select_query(evaluator, *projections):
        return exp.select(*[f'{p} AS "{p}"' for p in projections])

    expressions = d.parse(
        """
        MODEL (
            name test
        );

        @SELECT_QUERY(1, 2, 3)
        """
    )
    model = load_sql_based_model(expressions)
    assert_exp_eq(
        model.render_query(),
        """
        SELECT
          1 AS "1",
          2 AS "2",
          3 AS "3"
        """,
    )


def test_partition_key_is_missing_in_query():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name,
            kind INCREMENTAL_BY_TIME_RANGE(
              time_column a
            ),
            partitioned_by (b, c, d)
        );

        SELECT 1::int AS a, 2::int AS b;
    """
    )

    model = load_sql_based_model(expressions)
    with pytest.raises(ConfigError) as ex:
        model.validate_definition()
    assert "['c', 'd'] are missing" in str(ex.value)


def test_cluster_key_is_missing_in_query():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name,
            kind INCREMENTAL_BY_TIME_RANGE(
              time_column a
            ),
            clustered_by (b, c, d)
        );

        SELECT 1::int AS a, 2::int AS b;
    """
    )

    model = load_sql_based_model(expressions)
    with pytest.raises(ConfigError) as ex:
        model.validate_definition()
    assert "['c', 'd'] are missing" in str(ex.value)


def test_partition_key_and_select_star():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name,
            kind INCREMENTAL_BY_TIME_RANGE(
              time_column a
            ),
            partitioned_by (b, c, d)
        );

        SELECT * FROM tbl;
    """
    )

    load_sql_based_model(expressions)


def test_json_serde():
    expressions = parse(
        """
        MODEL (
            name test_model,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column ds
            ),
            owner test_owner,
            dialect spark,
            cron '@daily',
            storage_format parquet,
            partitioned_by a,
            physical_properties (
                key_a = 'value_a',
                'key_b' = 1,
                key_c = true,
                "key_d" = 2.0,
            ),
        );

        @DEF(key, 'value');

        SELECT a, ds FROM `tbl`
    """
    )

    model = load_sql_based_model(expressions)

    model_json = model.json()
    model_json_parsed = json.loads(model.json())
    assert model_json_parsed["kind"]["dialect"] == "spark"
    assert model_json_parsed["kind"]["time_column"]["column"] == "`ds`"
    assert model_json_parsed["partitioned_by"] == ["`a`"]

    deserialized_model = SqlModel.parse_raw(model_json)

    assert deserialized_model.dict() == model.dict()

    expressions = parse(
        """
        MODEL (
            name test_model,
            kind FULL,
            dialect duckdb,
        );

        SELECT
          x ~ y AS c
        """
    )

    model = load_sql_based_model(expressions)
    model_json = model.json()
    model_json_parsed = json.loads(model.json())

    assert (
        SqlModel.parse_obj(model_json_parsed).render_query().sql("duckdb")
        == 'SELECT REGEXP_MATCHES("x", "y") AS "c"'
    )


def test_scd_type_2_by_col_serde():
    expressions = parse(
        """
        MODEL(
            name test_model,
            KIND SCD_TYPE_2_BY_COLUMN (
                unique_key a,
                columns *,
            ),
            cron '@daily',
            owner 'reakman',
            grain a,
            dialect bigquery
        );

        SELECT a, ds FROM `tbl`
    """,
        default_dialect="bigquery",
    )

    model = load_sql_based_model(expressions)

    model_json = model.json()
    model_json_parsed = json.loads(model.json())
    assert model_json_parsed["kind"]["dialect"] == "bigquery"
    assert model_json_parsed["kind"]["unique_key"] == ["`a`"]
    assert model_json_parsed["kind"]["columns"] == "*"
    # Bigquery converts TIMESTAMP -> DATETIME
    assert model_json_parsed["kind"]["time_data_type"] == "DATETIME"

    deserialized_model = SqlModel.parse_raw(model_json)
    assert deserialized_model.dict() == model.dict()


def test_column_descriptions(sushi_context, assert_exp_eq):
    assert sushi_context.models[
        '"memory"."sushi"."customer_revenue_by_day"'
    ].column_descriptions == {
        "customer_id": "Customer id",
        "country code": "Customer country code, used for testing spaces",
        "revenue": "Revenue from orders made by this customer",
        "event_date": "Date",
    }

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind FULL,
        );

        SELECT
          -- this is the id column, used in ...
          id::int, -- primary key
          foo::int, -- bar
        FROM table
    """
    )
    model = load_sql_based_model(expressions, default_catalog="memory")

    assert_exp_eq(
        model.query,
        """
        SELECT
          -- this is the id column, used in ...
          id::int, -- primary key
          foo::int, -- bar
        FROM table
        """,
    )
    assert model.column_descriptions == {"id": "primary key", "foo": "bar"}


def test_model_jinja_macro_reference_extraction():
    @macro()
    def test_macro(**kwargs) -> None:
        pass

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name,
        );

        JINJA_STATEMENT_BEGIN;
        {{ test_macro() }}
        JINJA_END;

        SELECT 1 AS x;
    """
    )
    model = load_sql_based_model(expressions)
    assert "test_macro" in model.python_env


def test_model_pre_post_statements():
    macro.registry().pop("foo", None)

    @macro()
    def foo(**kwargs) -> None:
        pass

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name,
        );

        @foo();

        JINJA_STATEMENT_BEGIN;
        CREATE TABLE x{{ 1 + 1 }};
        JINJA_END;

        SELECT 1 AS x;

        @foo(bar='x', val=@this);

        DROP TABLE x2;
    """
    )
    model = load_sql_based_model(expressions)

    expected_pre = [
        *d.parse("@foo()"),
        d.jinja_statement("CREATE TABLE x{{ 1 + 1 }};"),
    ]
    assert model.pre_statements == expected_pre

    expected_post = d.parse("@foo(bar='x', val=@this); DROP TABLE x2;")
    assert model.post_statements == expected_post

    assert model.query == d.parse("SELECT 1 AS x")[0]

    @macro()
    def multiple_statements(evaluator, t1_value=exp.Literal.number(1)):
        return [f"CREATE TABLE t1 AS SELECT {t1_value} AS c", "CREATE TABLE t2 AS SELECT 2 AS c"]

    expressions = d.parse(
        """
        MODEL (name db.table);

        SELECT 1 AS col;

        @multiple_statements()
        """
    )
    model = load_sql_based_model(expressions)

    expected_post = d.parse(
        'CREATE TABLE "t1" AS SELECT 1 AS "c"; CREATE TABLE "t2" AS SELECT 2 AS "c"'
    )
    assert model.render_post_statements() == expected_post
    assert "exp" in model.python_env


@pytest.mark.parametrize("model_kind", ["FULL", "VIEW"])
def test_model_pre_post_statements_start_end_are_always_available(model_kind: str):
    macro.registry().pop("foo", None)
    macro.registry().pop("bar", None)

    @macro()
    def foo(evaluator: MacroEvaluator, start: str, end: str) -> str:
        return f"'{start}, {end}'"

    @macro()
    def bar(evaluator: MacroEvaluator, start: int, end: int) -> str:
        return f"'{start}, {end}'"

    expressions = d.parse(
        f"""
        MODEL (
            name db.table,
            kind {model_kind},
        );

        @foo(@start_ds, @end_ds);

        SELECT 1 AS x;

        @bar(@start_millis, @end_millis);
    """
    )
    model = load_sql_based_model(expressions)

    start = "2025-01-01"
    end = "2025-01-02"

    assert model.render_pre_statements(start=start, end=end) == [
        exp.Literal.string(f"{start}, {end}")
    ]
    assert model.render_post_statements(start=start, end=end) == [
        exp.Literal.string(f"{to_timestamp(start)}, {to_timestamp('2025-01-03') - 1}")
    ]


def test_seed_hydration():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              batch_size 100,
            )
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))
    assert model.is_hydrated
    assert not model.derived_columns_to_types

    column_hashes = model.column_hashes

    dehydrated_model = model.to_dehydrated()
    assert not dehydrated_model.is_hydrated
    assert dehydrated_model.column_hashes == column_hashes
    assert dehydrated_model.derived_columns_to_types == {
        "id": exp.DataType.build("bigint"),
        "name": exp.DataType.build("text"),
    }
    assert dehydrated_model.seed.content == ""

    hydrated_model = dehydrated_model.to_hydrated(model.seed.content)
    assert hydrated_model.is_hydrated
    assert hydrated_model.column_hashes == column_hashes
    assert hydrated_model.seed.content == model.seed.content
    assert hydrated_model.column_hashes_ is None


def test_seed():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              batch_size 100,
            )
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    assert isinstance(model.kind, SeedKind)
    assert model.kind.path == "../seeds/waiter_names.csv"
    assert model.kind.batch_size == 100
    assert model.seed is not None
    assert len(model.seed.content) > 0

    assert model.columns_to_types == {
        "id": exp.DataType.build("bigint"),
        "name": exp.DataType.build("text"),
    }

    assert model.batch_size is None


def test_seed_model_creation_error():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path 'gibberish',
            )
        );
    """
    )
    with pytest.raises(FileNotFoundError, match="No such file or directory"):
        load_sql_based_model(expressions)


def test_seed_provided_columns():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              batch_size 100,
            ),
            columns (
              id double,
              alias varchar
            )
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    assert isinstance(model.kind, SeedKind)
    assert model.kind.path == "../seeds/waiter_names.csv"
    assert model.kind.batch_size == 100
    assert model.seed is not None
    assert len(model.seed.content) > 0

    assert model.columns_to_types == {
        "id": exp.DataType.build("double"),
        "alias": exp.DataType.build("varchar"),
    }


def test_seed_case_sensitive_columns(tmp_path):
    model_csv_path = (tmp_path / "model.csv").absolute()

    with open(model_csv_path, "w", encoding="utf-8") as fd:
        fd.write(
            """camelCaseId,camelCaseBool,camelCaseString,normalisedCaseDate,camelCaseTimestamp
1,false,Alice,2022-01-01,2022-01-01
"""
        )

    expressions = d.parse(
        f"""
        MODEL (
            name db.seed,
            dialect postgres,
            kind SEED (
              path '{str(model_csv_path)}',
            ),
            columns (
              "camelCaseId" int,
              "camelCaseBool" boolean,
              "camelCaseString" text,
              "camelCaseTimestamp" timestamp
            )
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    assert isinstance(model.kind, SeedKind)
    assert model.seed is not None
    assert len(model.seed.content) > 0
    assert model.columns_to_types == {
        "camelCaseId": exp.DataType.build("int"),
        "camelCaseBool": exp.DataType.build("boolean"),
        "camelCaseString": exp.DataType.build("text"),
        "camelCaseTimestamp": exp.DataType.build("TIMESTAMP"),
    }
    df = next(model.render(context=None))

    assert df["camelCaseId"].dtype == "int64"
    assert df["camelCaseId"].iloc[0] == 1

    assert df["camelCaseBool"].dtype == "bool"
    assert not df["camelCaseBool"].iloc[0]

    assert df["camelCaseString"].dtype == "object"
    assert df["camelCaseString"].iloc[0] == "Alice"

    assert df["normalisedcasedate"].dtype == "object"
    assert df["normalisedcasedate"].iloc[0] == "2022-01-01"

    assert df["camelCaseTimestamp"].dtype == "datetime64[ns]"
    assert df["camelCaseTimestamp"].iloc[0] == pd.Timestamp("2022-01-01 00:00:00")


def test_seed_csv_settings():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              batch_size 100,
              csv_settings (
                quotechar = '''',
                escapechar = '\\',
                keep_default_na = false,
                na_values = (id = [1, '2', false, null], alias = ('foo'))
              ),
            ),
            columns (
              id double,
              alias varchar
            )
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    assert isinstance(model.kind, SeedKind)
    assert model.kind.csv_settings == CsvSettings(
        quotechar="'",
        escapechar="\\",
        na_values={"id": [1, "2", False, None], "alias": ["foo"]},
        keep_default_na=False,
    )
    assert model.kind.data_hash_values == [
        "SEED",
        "'",
        "\\",
        "{'id': [1, '2', False, None], 'alias': ['foo']}",
        "False",
    ]

    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              csv_settings (
                na_values = ('#N/A', 'other')
              ),
            ),
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    assert isinstance(model.kind, SeedKind)
    assert model.kind.csv_settings == CsvSettings(na_values=["#N/A", "other"])
    assert model.kind.data_hash_values == ["SEED", "['#N/A', 'other']"]


def test_seed_marker_substitution():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '$root/seeds/waiter_names.csv',
              batch_size 100,
            )
        );
    """
    )

    model = load_sql_based_model(
        expressions,
        path=Path("./examples/sushi/models/test_model.sql"),
        module_path=Path("./examples/sushi"),
    )

    assert isinstance(model.kind, SeedKind)
    assert model.kind.path == str(Path("examples/sushi/seeds/waiter_names.csv"))
    assert model.seed is not None
    assert len(model.seed.content) > 0


def test_seed_pre_post_statements():
    macro.registry().pop("bar", None)

    @macro()
    def bar(**kwargs) -> None:
        pass

    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              batch_size 100,
            )
        );

        @bar();

        JINJA_STATEMENT_BEGIN;
        CREATE TABLE x{{ 1 + 1 }};
        JINJA_END;

        @INSERT_SEED();

        @bar(foo='x', val=@this);

        DROP TABLE x2;
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    expected_pre = [
        *d.parse("@bar()"),
        d.jinja_statement("CREATE TABLE x{{ 1 + 1 }};"),
    ]
    assert model.pre_statements == expected_pre

    expected_post = [
        *d.parse("@bar(foo='x', val=@this)"),
        *d.parse("DROP TABLE x2;"),
    ]
    assert model.post_statements == expected_post


def test_seed_pre_statements_only():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              batch_size 100,
            )
        );

        JINJA_STATEMENT_BEGIN;
        CREATE TABLE x{{ 1 + 1 }};
        JINJA_END;

        DROP TABLE x2;
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    expected_pre = [
        d.jinja_statement("CREATE TABLE x{{ 1 + 1 }};"),
        *d.parse("DROP TABLE x2;"),
    ]
    assert model.pre_statements == expected_pre
    assert not model.post_statements


def test_seed_on_virtual_update_statements():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              batch_size 100,
            )
        );

        JINJA_STATEMENT_BEGIN;
        CREATE TABLE x{{ 1 + 1 }};
        JINJA_END;

        ON_VIRTUAL_UPDATE_BEGIN;
        JINJA_STATEMENT_BEGIN;
        GRANT SELECT ON VIEW {{ this_model }} TO ROLE dev_role;
        JINJA_END;
        DROP TABLE x2;
        ON_VIRTUAL_UPDATE_END;

    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    assert model.pre_statements == [d.jinja_statement("CREATE TABLE x{{ 1 + 1 }};")]
    assert model.on_virtual_update == [
        d.jinja_statement("GRANT SELECT ON VIEW {{ this_model }} TO ROLE dev_role;"),
        *d.parse("DROP TABLE x2;"),
    ]


def test_seed_model_custom_types(tmp_path):
    model_csv_path = (tmp_path / "model.csv").absolute()

    with open(model_csv_path, "w", encoding="utf-8") as fd:
        fd.write(
            """key,ds_date,ds_timestamp,b_a,b_b,i,i_str,empty_date
123,2022-01-01,2022-01-01,false,0,321,321,
"""
        )

    model = create_seed_model(
        "test_db.test_model",
        SeedKind(path=str(model_csv_path)),
        columns={
            "key": "string",
            "ds_date": "date",
            "ds_timestamp": "timestamp",
            "b_a": "boolean",
            "b_b": "boolean",
            "i": "int",
            "i_str": "text",
            "empty_date": "date",
        },
    )

    df = next(model.render(context=None))

    assert df["ds_date"].dtype == "object"
    assert df["ds_date"].iloc[0] == date(2022, 1, 1)

    assert df["ds_timestamp"].dtype == "datetime64[ns]"
    assert df["ds_timestamp"].iloc[0] == pd.Timestamp("2022-01-01 00:00:00")

    assert df["key"].dtype == "object"
    assert df["key"].iloc[0] == "123"

    assert df["b_a"].dtype == "bool"
    assert not df["b_a"].iloc[0]

    assert df["b_b"].dtype == "bool"
    assert not df["b_b"].iloc[0]

    assert df["i"].dtype == "int64"
    assert df["i"].iloc[0] == 321

    assert df["i_str"].dtype == "object"
    assert df["i_str"].iloc[0] == "321"

    assert df["empty_date"].dtype == "object"
    assert df["empty_date"].iloc[0] is None


def test_seed_with_special_characters_in_column(tmp_path, assert_exp_eq):
    config = Config(model_defaults=ModelDefaultsConfig(dialect="duckdb"))
    context = Context(config=config)

    model_csv_path = (tmp_path / "model.csv").absolute()
    with open(model_csv_path, "w", encoding="utf-8") as fd:
        fd.write("col.\tcol!@#$\n123\tfoo")

    expressions = d.parse(
        f"""
        MODEL (
            name memory.test_db.test_model,
            kind SEED (
              path '{model_csv_path}',
              csv_settings (
                delimiter = '\\t'
              )
            ),
        );
    """
    )

    context.upsert_model(load_sql_based_model(expressions))
    assert_exp_eq(
        context.render("memory.test_db.test_model").sql(),
        "SELECT "
        'CAST("col." AS BIGINT) AS "col.", '
        'CAST("col!@#$" AS TEXT) AS "col!@#$" '
        """FROM (VALUES (123, 'foo')) AS t("col.", "col!@#$")""",
    )


def test_python_model_jinja_pre_post_statements():
    macros = """
    {% macro test_macro(v) %}{{ v }}{% endmacro %}
    {% macro extra_macro(v) %}{{ v + 1 }}{% endmacro %}
    """

    jinja_macros = JinjaMacroRegistry()
    jinja_macros.add_macros(MacroExtractor().extract(macros))

    @model(
        "db.test_model",
        kind="full",
        columns={"id": "string", "name": "string"},
        pre_statements=[
            "JINJA_STATEMENT_BEGIN;\n{% set table_name = 'x' %}\nCREATE OR REPLACE TABLE {{table_name}}{{ 1 + 1 }};\nJINJA_END;"
        ],
        post_statements=[
            "JINJA_STATEMENT_BEGIN;\nCREATE INDEX {{test_macro('idx')}} ON db.test_model(id);\nJINJA_END;",
            parse_one("DROP TABLE x2;"),
        ],
    )
    def model_with_statements(context, **kwargs):
        return pd.DataFrame(
            [
                {
                    "id": context.var("1"),
                    "name": context.var("var"),
                }
            ]
        )

    python_model = model.get_registry()["db.test_model"].model(
        module_path=Path("."), path=Path("."), dialect="duckdb", jinja_macros=jinja_macros
    )

    assert len(jinja_macros.root_macros) == 2
    assert len(python_model.jinja_macros.root_macros) == 1
    assert "test_macro" in python_model.jinja_macros.root_macros
    assert "extra_macro" not in python_model.jinja_macros.root_macros

    expected_pre = [
        d.jinja_statement(
            "{% set table_name = 'x' %}\nCREATE OR REPLACE TABLE {{table_name}}{{ 1 + 1 }};"
        ),
    ]
    assert python_model.pre_statements == expected_pre
    assert python_model.render_pre_statements()[0].sql() == 'CREATE OR REPLACE TABLE "x2"'

    expected_post = [
        d.jinja_statement("CREATE INDEX {{test_macro('idx')}} ON db.test_model(id);"),
        *d.parse("DROP TABLE x2;"),
    ]
    assert python_model.post_statements == expected_post
    assert (
        python_model.render_post_statements()[0].sql()
        == 'CREATE INDEX "idx" ON "db"."test_model"("id" NULLS LAST)'
    )
    assert python_model.render_post_statements()[1].sql() == 'DROP TABLE "x2"'


def test_audits():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            audits (
                audit_a,
                audit_b(key='value'),
                audit_c(key=@start_ds)
            ),
            tags (foo)
        );
        SELECT 1, ds;
    """
    )

    audit_definitions = {
        audit_name: load_audit(
            d.parse(f"AUDIT (name {audit_name}); SELECT 1 WHERE FALSE"), dialect="duckdb"
        )
        for audit_name in ("audit_a", "audit_b", "audit_c")
    }

    model = load_sql_based_model(
        expressions,
        path=Path("./examples/sushi/models/test_model.sql"),
        audit_definitions=audit_definitions,
    )
    assert model.audits == [
        ("audit_a", {}),
        ("audit_b", {"key": exp.Literal.string("value")}),
        ("audit_c", {"key": d.MacroVar(this="start_ds")}),
    ]
    assert model.tags == ["foo"]


def test_enable_audits_from_model_defaults():
    expressions = d.parse(
        """
        MODEL (
            name db.audit_model,
        );
        SELECT 1 as id;

        AUDIT (
    name assert_positive_order_ids,
    );
    SELECT *
    FROM @this_model
    WHERE
    id < 0;
    """
    )

    model_defaults = ModelDefaultsConfig(dialect="duckdb", audits=["assert_positive_order_ids"])

    model = load_sql_based_model(
        expressions,
        path=Path("./examples/sushi/models/test_model.sql"),
        defaults=model_defaults.dict(),
    )

    assert len(model.audits) == 1

    config = Config(model_defaults=model_defaults)
    assert config.model_defaults.audits[0] == ("assert_positive_order_ids", {}) == model.audits[0]

    audits_with_args = model.audits_with_args
    assert len(audits_with_args) == 1
    audit, args = audits_with_args[0]
    assert type(audit) == ModelAudit
    assert args == {}
    assert audit.query.sql() == "SELECT * FROM @this_model WHERE id < 0"


def test_description(sushi_context):
    assert sushi_context.models['"memory"."sushi"."orders"'].description == "Table of sushi orders."


def test_model_defaults_statements_merge():
    model_defaults = ModelDefaultsConfig(
        dialect="duckdb",
        pre_statements=[
            "SET enable_progress_bar = true",
            "CREATE TEMP TABLE default_temp AS SELECT 1",
        ],
        post_statements=[
            "DROP TABLE IF EXISTS default_temp",
            "grant select on @this_model to group reporter",
        ],
        on_virtual_update=["ANALYZE"],
    )

    # Create a model with its own statements as well
    expressions = parse(
        """
    MODEL (
        name test_model,
        kind FULL
    );

    CREATE TEMP TABLE model_temp AS SELECT 2;

    SELECT * FROM test_table;

    DROP TABLE IF EXISTS model_temp;

    ON_VIRTUAL_UPDATE_BEGIN;
    UPDATE stats_table SET last_update = CURRENT_TIMESTAMP;
    ON_VIRTUAL_UPDATE_END;
    """
    )

    model = load_sql_based_model(
        expressions,
        path=Path("./test_model.sql"),
        defaults=model_defaults.dict(),
    )

    # Check that pre_statements contains both default and model-specific statements
    assert len(model.pre_statements) == 3
    assert model.pre_statements[0].sql() == "SET enable_progress_bar = TRUE"
    assert model.pre_statements[1].sql() == "CREATE TEMPORARY TABLE default_temp AS SELECT 1"
    assert model.pre_statements[2].sql() == "CREATE TEMPORARY TABLE model_temp AS SELECT 2"

    # Check that post_statements contains both default and model-specific statements
    assert len(model.post_statements) == 3
    assert model.post_statements[0].sql() == "DROP TABLE IF EXISTS default_temp"
    assert model.post_statements[1].sql() == "GRANT SELECT ON @this_model TO GROUP reporter"
    assert model.post_statements[2].sql() == "DROP TABLE IF EXISTS model_temp"

    # Check that the query is rendered correctly with @this_model resolved to table name
    assert (
        model.render_post_statements()[1].sql()
        == 'GRANT SELECT ON "test_model" TO GROUP "reporter"'
    )

    # Check that on_virtual_update contains both default and model-specific statements
    assert len(model.on_virtual_update) == 2
    assert model.on_virtual_update[0].sql() == "ANALYZE"
    assert (
        model.on_virtual_update[1].sql()
        == "UPDATE stats_table SET last_update = CURRENT_TIMESTAMP()"
    )


def test_model_defaults_statements_integration():
    config = Config(
        model_defaults=ModelDefaultsConfig(
            dialect="postgres",
            pre_statements=["SET memory_limit = '10GB'"],
            post_statements=["VACUUM ANALYZE"],
            on_virtual_update=["GRANT SELECT ON @this_model TO GROUP public"],
        )
    )

    expressions = parse(
        """
    MODEL (
        name test_model,
        kind FULL
    );

    SELECT * FROM source_table;
    """
    )

    model = load_sql_based_model(
        expressions,
        path=Path("./test_model.sql"),
        defaults=config.model_defaults.dict(),
    )

    # Verify defaults were applied
    assert len(model.pre_statements) == 1
    assert model.pre_statements[0].sql() == "SET memory_limit = '10GB'"

    assert len(model.post_statements) == 1
    assert isinstance(model.post_statements[0], exp.Command)

    assert len(model.on_virtual_update) == 1
    assert model.on_virtual_update[0].sql() == "GRANT SELECT ON @this_model TO GROUP public"
    assert (
        model.render_on_virtual_update()[0].sql()
        == 'GRANT SELECT ON "test_model" TO GROUP "public"'
    )


def test_render_definition():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            owner owner_name,
            cron_tz 'America/Los_Angeles',
            dialect spark,
            kind INCREMENTAL_BY_TIME_RANGE (
                time_column (`a`, 'yyyymmdd'),
                partition_by_time_column TRUE,
                forward_only FALSE,
                disable_restatement FALSE,
                on_destructive_change 'ERROR',
                on_additive_change 'ALLOW'
            ),
            storage_format iceberg,
            partitioned_by `a`,
            grains (
                [a, b],
                c
            ),
            references (
                [a, b],
                c
            ),
        );

        @DEF(x, 1);
        CACHE TABLE x AS SELECT 1;
        ADD JAR 's3://my_jar.jar';

        SELECT
            1::int AS a,
            CAST(2 AS double) AS b,
            c::bool,
            1::int AS d, -- d
            CAST(2 AS double) AS e, --e
            f::bool, --f
            @test_macro(1),
        FROM
            db.other_table t1
            LEFT JOIN
            db.table t2
            ON
                t1.a = t2.a;

        @IF( @runtime_stage = 'creating', create index db_table_idx on db.table(a) );
    """
    )

    model = load_sql_based_model(
        expressions,
        python_env={
            "test_macro": Executable(payload="def test_macro(evaluator, v):\n    return v"),
        },
        default_catalog="catalog",
    )

    # Should not include the macro implementation.
    assert d.format_model_expressions(
        model.render_definition(include_python=False)
    ) == d.format_model_expressions(expressions)

    # Should include the macro implementation.
    assert "def test_macro(evaluator, v):" in d.format_model_expressions(model.render_definition())


def test_render_definition_with_defaults():
    query = """
        SELECT
            1::int AS a,
            CAST(2 AS double) AS b,
        FROM
            db.other_table t1
            LEFT JOIN
            db.table t2
            ON
                t1.a = t2.a
        """

    expressions = d.parse(
        f"""
        MODEL (
            name db.table,
            owner owner_name,
            dialect spark,
            kind VIEW,
        );

        {query}
        """
    )

    model = load_sql_based_model(
        expressions,
        default_catalog="catalog",
    )

    expected_expressions = d.parse(
        f"""
        MODEL (
            name db.table,
            owner owner_name,
            cron '@daily',
            dialect spark,
            kind VIEW (
                materialized FALSE
            ),
            virtual_environment_mode 'full'
        );

        {query}
        """
    )

    # Should not include the macro implementation.
    assert d.format_model_expressions(
        model.render_definition(include_python=False, include_defaults=True)
    ) == d.format_model_expressions(expected_expressions)


def test_render_definition_partitioned_by():
    # no parenthesis in definition, no parenthesis when rendered
    model = load_sql_based_model(
        d.parse(
            f"""
        MODEL (
            name db.table,
            kind FULL,
            partitioned_by a
        );

        select 1 as a;
        """
        )
    )

    assert model.partitioned_by == [exp.column("a", quoted=True)]
    assert (
        model.render_definition()[0].sql(pretty=True)
        == """MODEL (
  name db.table,
  kind FULL,
  partitioned_by "a"
)"""
    )

    # single column wrapped in parenthesis in defintion, no parenthesis in rendered
    model = load_sql_based_model(
        d.parse(
            f"""
        MODEL (
            name db.table,
            kind FULL,
            partitioned_by (a)
        );

        select 1 as a;
        """
        )
    )

    assert model.partitioned_by == [exp.column("a", quoted=True)]
    assert (
        model.render_definition()[0].sql(pretty=True)
        == """MODEL (
  name db.table,
  kind FULL,
  partitioned_by "a"
)"""
    )

    # multiple columns wrapped in parenthesis in definition, parenthesis in rendered
    model = load_sql_based_model(
        d.parse(
            f"""
        MODEL (
            name db.table,
            kind FULL,
            partitioned_by (a, b)
        );

        select 1 as a, 2 as b;
        """
        )
    )

    assert model.partitioned_by == [exp.column("a", quoted=True), exp.column("b", quoted=True)]
    assert (
        model.render_definition()[0].sql(pretty=True)
        == """MODEL (
  name db.table,
  kind FULL,
  partitioned_by ("a", "b")
)"""
    )

    # multiple columns not wrapped in parenthesis in the definition is an error
    with pytest.raises(ParseError, match=r"keyword: 'value' missing"):
        load_sql_based_model(
            d.parse(
                f"""
            MODEL (
                name db.table,
                kind FULL,
                partitioned_by a, b
            );

            select 1 as a, 2 as b;
            """
            )
        )

    # Iceberg transforms / functions
    model = load_sql_based_model(
        d.parse(
            f"""
        MODEL (
            name db.table,
            kind FULL,
            partitioned_by (day(a), truncate(b, 4), bucket(c, 3))
        );

        select 1 as a, 2 as b, 3 as c;
        """
        ),
        dialect="trino",
    )

    assert model.partitioned_by == [
        exp.Day(this=exp.column("a", quoted=True)),
        exp.PartitionByTruncate(
            this=exp.column("b", quoted=True), expression=exp.Literal.number(4)
        ),
        exp.PartitionedByBucket(
            this=exp.column("c", quoted=True), expression=exp.Literal.number(3)
        ),
    ]
    assert (
        model.render_definition()[0].sql(pretty=True)
        == """MODEL (
  name db.table,
  dialect trino,
  kind FULL,
  partitioned_by (DAY("a"), TRUNCATE("b", 4), BUCKET("c", 3))
)"""
    )


def test_render_definition_with_virtual_update_statements():
    # model has virtual update statements
    model = load_sql_based_model(
        d.parse(
            f"""
        MODEL (
            name db.table,
            kind FULL
        );

        select 1 as a;

        ON_VIRTUAL_UPDATE_BEGIN;
        GRANT SELECT ON VIEW @this_model TO ROLE role_name
        ON_VIRTUAL_UPDATE_END;
        """
        )
    )

    assert model.on_virtual_update == [
        exp.Grant(
            privileges=[exp.GrantPrivilege(this=exp.Var(this="SELECT"))],
            kind="VIEW",
            securable=exp.Table(this=d.MacroVar(this="this_model")),
            principals=[
                exp.GrantPrincipal(this=exp.Identifier(this="role_name", quoted=False), kind="ROLE")
            ],
        )
    ]
    assert (
        model.render_definition()[-1].sql(pretty=True)
        == """ON_VIRTUAL_UPDATE_BEGIN;
GRANT SELECT ON VIEW @this_model TO ROLE role_name;
ON_VIRTUAL_UPDATE_END;"""
    )


def test_cron():
    daily = _Node(name="x", cron="@daily")
    assert to_datetime(daily.cron_prev("2020-01-01")) == to_datetime("2019-12-31")
    assert to_datetime(daily.cron_floor("2020-01-01")) == to_datetime("2020-01-01")
    assert to_timestamp(daily.cron_floor("2020-01-01 10:00:00")) == to_timestamp("2020-01-01")
    assert to_timestamp(daily.cron_next("2020-01-01 10:00:00")) == to_timestamp("2020-01-02")
    interval = daily.interval_unit
    assert to_datetime(interval.cron_prev("2020-01-01")) == to_datetime("2019-12-31")
    assert to_datetime(interval.cron_floor("2020-01-01")) == to_datetime("2020-01-01")
    assert to_timestamp(interval.cron_floor("2020-01-01 10:00:00")) == to_timestamp("2020-01-01")
    assert to_timestamp(interval.cron_next("2020-01-01 10:00:00")) == to_timestamp("2020-01-02")

    offset = _Node(name="x", cron="1 0 * * *")
    assert to_datetime(offset.cron_prev("2020-01-01")) == to_datetime("2019-12-31 00:01")
    assert to_datetime(offset.cron_floor("2020-01-01")) == to_datetime("2019-12-31 00:01")
    assert to_timestamp(offset.cron_floor("2020-01-01 10:00:00")) == to_timestamp(
        "2020-01-01 00:01"
    )
    assert to_timestamp(offset.cron_next("2020-01-01 10:00:00")) == to_timestamp("2020-01-02 00:01")
    interval = offset.interval_unit
    assert to_datetime(interval.cron_prev("2020-01-01")) == to_datetime("2019-12-31")
    assert to_datetime(interval.cron_floor("2020-01-01")) == to_datetime("2020-01-01")
    assert to_timestamp(interval.cron_floor("2020-01-01 10:00:00")) == to_timestamp("2020-01-01")
    assert to_timestamp(interval.cron_next("2020-01-01 10:00:00")) == to_timestamp("2020-01-02")

    hourly = _Node(name="x", cron="1 * * * *")
    assert to_timestamp(hourly.cron_prev("2020-01-01 10:00:00")) == to_timestamp(
        "2020-01-01 09:01:00"
    )
    assert to_timestamp(hourly.cron_prev("2020-01-01 10:02:00")) == to_timestamp(
        "2020-01-01 10:01:00"
    )
    assert to_timestamp(hourly.cron_floor("2020-01-01 10:01:00")) == to_timestamp(
        "2020-01-01 10:01:00"
    )
    interval = hourly.interval_unit
    assert to_timestamp(interval.cron_prev("2020-01-01 10:00:00")) == to_timestamp(
        "2020-01-01 09:00:00"
    )
    assert to_timestamp(interval.cron_prev("2020-01-01 10:01:00")) == to_timestamp(
        "2020-01-01 10:00:00"
    )
    assert to_timestamp(interval.cron_floor("2020-01-01 10:01:00")) == to_timestamp(
        "2020-01-01 10:00:00"
    )

    monthly = _Node(name="x", cron="0 0 2 * *")
    assert to_timestamp(monthly.cron_prev("2020-01-01 00:00:00")) == to_timestamp(
        "2019-12-02 00:00:00"
    )
    assert to_timestamp(monthly.cron_prev("2020-02-01 00:00:00")) == to_timestamp(
        "2020-01-02 00:00:00"
    )
    assert to_timestamp(monthly.cron_next("2020-01-01 00:00:00")) == to_timestamp(
        "2020-01-02 00:00:00"
    )
    assert to_timestamp(monthly.cron_next("2020-01-02 00:00:00")) == to_timestamp(
        "2020-02-02 00:00:00"
    )
    assert to_timestamp(monthly.cron_floor("2020-01-17 00:00:00")) == to_timestamp(
        "2020-01-02 00:00:00"
    )
    interval = monthly.interval_unit
    assert to_timestamp(interval.cron_prev("2020-01-01 00:00:00")) == to_timestamp(
        "2019-12-01 00:00:00"
    )
    assert to_timestamp(interval.cron_prev("2020-02-01 00:00:00")) == to_timestamp(
        "2020-01-01 00:00:00"
    )
    assert to_timestamp(interval.cron_next("2020-01-01 00:00:00")) == to_timestamp(
        "2020-02-01 00:00:00"
    )
    assert to_timestamp(interval.cron_floor("2020-01-17 00:00:00")) == to_timestamp(
        "2020-01-01 00:00:00"
    )

    yearly = ModelMeta(name="x", cron="0 0 1 2 *")
    assert to_timestamp(yearly.cron_prev("2020-01-01 00:00:00")) == to_timestamp(
        "2019-02-01 00:00:00"
    )
    assert to_timestamp(yearly.cron_next("2020-01-01 00:00:00")) == to_timestamp(
        "2020-02-01 00:00:00"
    )
    assert to_timestamp(yearly.cron_floor("2020-12-10 00:00:00")) == to_timestamp(
        "2020-02-01 00:00:00"
    )
    interval = yearly.interval_unit
    assert to_timestamp(interval.cron_prev("2020-01-01 00:00:00")) == to_timestamp(
        "2019-01-01 00:00:00"
    )
    assert to_timestamp(interval.cron_next("2020-01-01 00:00:00")) == to_timestamp(
        "2021-01-01 00:00:00"
    )
    assert to_timestamp(interval.cron_floor("2020-12-10 00:00:00")) == to_timestamp(
        "2020-01-01 00:00:00"
    )


def test_lookback():
    model = ModelMeta(
        name="x", cron="@hourly", kind=IncrementalByTimeRangeKind(time_column="ts", lookback=2)
    )
    assert to_timestamp(model.lookback_start("Jan 8 2020 04:00:00")) == to_timestamp(
        "Jan 8 2020 02:00:00"
    )

    model = ModelMeta(
        name="x", cron="@daily", kind=IncrementalByTimeRangeKind(time_column="ds", lookback=2)
    )
    assert to_timestamp(model.lookback_start("Jan 8 2020")) == to_timestamp("Jan 6 2020")

    model = ModelMeta(
        name="x", cron="0 0 1 * *", kind=IncrementalByTimeRangeKind(time_column="ds", lookback=2)
    )
    assert to_timestamp(model.lookback_start("April 1 2020")) == to_timestamp("Feb 1 2020")

    model = ModelMeta(
        name="x", cron="0 0 1 1 *", kind=IncrementalByTimeRangeKind(time_column="ds", lookback=2)
    )
    assert to_timestamp(model.lookback_start("Jan 1 2020")) == to_timestamp("Jan 1 2018")


def test_render_query(assert_exp_eq, sushi_context):
    model = SqlModel(
        name="test",
        cron="1 0 * * *",
        kind=IncrementalByTimeRangeKind(time_column=TimeColumn(column="y")),
        query=d.parse_one(
            """
        SELECT y
        FROM x
        WHERE
          y BETWEEN @start_date and @end_date AND
          y BETWEEN @start_ds and @end_ds
        """
        ),
    )
    assert_exp_eq(
        model.render_query(start="2020-10-28", end="2020-10-28"),
        """
        SELECT
          "y" AS "y"
        FROM "x" AS "x"
        WHERE
          "y" BETWEEN DATE_STR_TO_DATE('2020-10-28') AND DATE_STR_TO_DATE('2020-10-28')
          AND "y" BETWEEN '2020-10-28' AND '2020-10-28'
        """,
    )
    assert_exp_eq(
        model.render_query(start="2020-10-28", end="2020-10-28", table_mapping={"x": "x_mapped"}),
        """
        SELECT
          "y" AS "y"
        FROM "x_mapped" AS "x"
        WHERE
          "y" BETWEEN DATE_STR_TO_DATE('2020-10-28') AND DATE_STR_TO_DATE('2020-10-28')
          AND "y" BETWEEN '2020-10-28' AND '2020-10-28'
        """,
    )

    expressions = d.parse(
        """
        MODEL (
          name dummy.model,
          kind FULL,
          dialect postgres
        );

        SELECT COUNT(DISTINCT a) FILTER (WHERE b > 0) AS c FROM x
        """
    )
    model = load_sql_based_model(expressions, dialect="postgres")
    assert_exp_eq(
        model.render_query(),
        'SELECT COUNT(DISTINCT "a") FILTER (WHERE "b" > 0) AS "c" FROM "x" AS "x"',
    )

    assert_exp_eq(
        sushi_context.models['"memory"."sushi"."waiters"'].render_query().sql(),
        """
        SELECT DISTINCT
          CAST("o"."waiter_id" AS INT) AS "waiter_id",
          CAST("o"."event_date" AS DATE) AS "event_date"
        FROM "memory"."sushi"."orders" AS "o"
        WHERE
          "o"."event_date" <= CAST('1970-01-01' AS DATE) AND "o"."event_date" >= CAST('1970-01-01' AS DATE)
        """,
    )

    expressions = d.parse(
        """
        MODEL (
          name dummy.model,
          kind FULL,
          dialect duckdb
        );

        @DEF(x, ['1', '2', '3']);

        SELECT @x AS "x"
        """
    )
    model = load_sql_based_model(expressions, dialect="duckdb")
    assert model.render_query().sql("duckdb") == '''SELECT ['1', '2', '3'] AS "x"'''

    expressions = d.parse(
        """
        MODEL (
          name dummy.model,
          kind FULL
        );

        @DEF(area, r -> pi() * r * r);

        SELECT route, centroid, @area(route_radius) AS area
        """
    )
    model = load_sql_based_model(expressions)
    assert (
        model.render_query().sql()
        == 'SELECT "route" AS "route", "centroid" AS "centroid", PI() * "route_radius" * "route_radius" AS "area"'
    )

    expressions = d.parse(
        """
        MODEL (
          name dummy.model,
          kind FULL
        );

        @DEF(area, r -> pi() * r * r);
        @DEF(container_volume, (r, h) -> @area(@r) * h);

        SELECT container_id, @container_volume((cont_di / 2), cont_hi) AS area
        """
    )
    model = load_sql_based_model(expressions)
    assert (
        model.render_query().sql()
        == 'SELECT "container_id" AS "container_id", PI() * ("cont_di" / 2) * ("cont_di" / 2) * "cont_hi" AS "area"'
    )

    expressions = d.parse(
        """
        MODEL (
          name dummy.model,
          kind FULL
        );

        @DEF(times2, x -> x * 2);

        SELECT @times4(10) AS "i dont exist"
        """
    )
    model = load_sql_based_model(expressions)
    with pytest.raises(SQLMeshError, match=r"Macro 'times4' does not exist.*"):
        model.render_query()

    assert_exp_eq(
        sushi_context.models['"memory"."sushi"."top_waiters"'].render_query().sql(),
        """
        WITH "test_macros" AS (
          SELECT
            2 AS "lit_two",
            "waiter_revenue_by_day"."revenue" * 2.0 AS "sql_exp",
            CAST("waiter_revenue_by_day"."revenue" AS TEXT) AS "sql_lit"
          FROM "memory"."sushi"."waiter_revenue_by_day" AS "waiter_revenue_by_day"
        )
        SELECT
          CAST("waiter_revenue_by_day"."waiter_id" AS INT) AS "waiter_id",
          CAST("waiter_revenue_by_day"."revenue" AS DOUBLE) AS "revenue"
        FROM "memory"."sushi"."waiter_revenue_by_day" AS "waiter_revenue_by_day"
        WHERE
          "waiter_revenue_by_day"."event_date" = (
            SELECT
              MAX("waiter_revenue_by_day"."event_date") AS "_col_0"
            FROM "memory"."sushi"."waiter_revenue_by_day" AS "waiter_revenue_by_day"
          )
        ORDER BY
          "revenue" DESC
        LIMIT 10
        """,
    )


def test_time_column():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column ds
            )
        );

        SELECT col::text, ds::text
    """
    )
    model = load_sql_based_model(expressions)
    assert model.time_column.column == exp.to_column("ds", quoted=True)
    assert model.time_column.format == "%Y-%m-%d"
    assert model.time_column.expression == parse_one("(\"ds\", '%Y-%m-%d')")

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column (ds)
            )
        );

        SELECT col::text, ds::text
    """
    )
    model = load_sql_based_model(expressions)
    assert model.time_column.column == exp.to_column("ds", quoted=True)
    assert model.time_column.format == "%Y-%m-%d"
    assert model.time_column.expression == d.parse_one("(\"ds\", '%Y-%m-%d')")

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect 'hive',
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column (ds, 'yyyy-MM'),
            )
        );

        SELECT col::text, ds::text
    """
    )
    model = load_sql_based_model(expressions)
    assert model.time_column.column == exp.to_column("ds", quoted=True)
    assert model.time_column.format == "%Y-%m"
    assert model.time_column.expression == d.parse_one("(\"ds\", '%Y-%m')")


def test_default_time_column():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column ds
            )
        );

        SELECT col::text, ds::text
    """
    )
    model = load_sql_based_model(expressions, time_column_format="%Y")
    assert model.time_column.format == "%Y"

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column (ds, "%Y")
            )
        );

        SELECT col::text, ds::text
    """
    )
    model = load_sql_based_model(expressions, time_column_format="%m")
    assert model.time_column.format == "%Y"

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect hive,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column (ds, "dd")
            )
        );

        SELECT col::text, ds::text
    """
    )
    model = load_sql_based_model(expressions, dialect="duckdb", time_column_format="%Y")
    assert model.time_column.format == "%d"


def test_convert_to_time_column():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column (ds)
            )
        );

        SELECT ds::text
    """
    )
    model = load_sql_based_model(expressions)
    assert model.convert_to_time_column("2022-01-01") == d.parse_one("'2022-01-01'")
    assert model.convert_to_time_column(to_datetime("2022-01-01")) == d.parse_one("'2022-01-01'")

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column (ds, '%d/%m/%Y')
            )
        );

        SELECT ds::text
    """
    )
    model = load_sql_based_model(expressions)
    assert model.convert_to_time_column("2022-01-01") == d.parse_one("'01/01/2022'")

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column (di, '%Y%m%d')
            )
        );

        SELECT di::int
    """
    )
    model = load_sql_based_model(expressions)
    assert model.convert_to_time_column("2022-01-01") == d.parse_one("20220101")

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column (ds, '%Y%m%d')
            )
        );

        SELECT ds::date
    """
    )
    model = load_sql_based_model(expressions)
    assert model.convert_to_time_column("2022-01-01") == d.parse_one("CAST('2022-01-01' AS DATE)")

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column ds
            )
        );

        SELECT ds::timestamp
    """
    )
    model = load_sql_based_model(expressions)
    assert model.convert_to_time_column("2022-01-01") == d.parse_one(
        "CAST('2022-01-01 00:00:00' AS TIMESTAMP)"
    )


def test_parse(assert_exp_eq):
    expressions = d.parse(
        """
        MODEL (
          name sushi.items,
          kind INCREMENTAL_BY_TIME_RANGE(
            time_column ds
          ),
          dialect '',
        );

        JINJA_QUERY_BEGIN;

        SELECT
          id::INT AS id,
          ds
        FROM x
        WHERE ds BETWEEN '{{ start_ds }}' AND @end_ds;

        JINJA_END;
    """
    )
    model = load_sql_based_model(expressions, dialect="hive")
    assert model.columns_to_types == {
        "ds": exp.DataType.build("unknown"),
        "id": exp.DataType.build("int"),
    }
    assert not model.annotated
    assert model.dialect == ""
    assert isinstance(model.query, d.JinjaQuery)
    assert isinstance(SqlModel.parse_raw(model.json()).query, d.JinjaQuery)
    assert_exp_eq(
        model.render_query(),
        """
      SELECT
        CAST("id" AS INT) AS "id",
        "ds" AS "ds"
      FROM "x" AS "x"
      WHERE
        "ds" BETWEEN '1970-01-01' AND '1970-01-01'
    """,
    )


CONST = "bar"


def test_python_model(assert_exp_eq) -> None:
    from functools import reduce

    @model(
        name="my_model",
        kind="full",
        columns={'"COL"': "int"},
        pre_statements=["CACHE TABLE x AS SELECT 1;"],
        post_statements=["DROP TABLE x;"],
        enabled=True,
    )
    def my_model(context, **kwargs):
        context.resolve_table("foo")
        context.resolve_table(model_name=CONST + ".baz")

        # This checks that built-in functions are serialized properly
        a = reduce(lambda x, y: x + y, [1, 2, 3, 4])  # noqa: F841

    m = model.get_registry()["my_model"].model(
        module_path=Path("."),
        path=Path("."),
        dialect="duckdb,normalization_strategy=LOWERCASE",
    )

    assert list(m.pre_statements) == [
        d.parse_one("CACHE TABLE x AS SELECT 1"),
    ]
    assert list(m.post_statements) == [
        d.parse_one("DROP TABLE x"),
    ]
    assert m.enabled
    assert m.dialect == "duckdb,normalization_strategy=lowercase"
    assert m.depends_on == {'"foo"', '"bar"."baz"'}
    assert m.columns_to_types == {"COL": exp.DataType.build("int")}
    assert_exp_eq(
        m.ctas_query(),
        """
SELECT
  CAST(NULL AS INT) AS "COL"
FROM (VALUES
  (1)) AS t(dummy)
WHERE
  FALSE
LIMIT 0
""",
    )


def test_python_model_depends_on() -> None:
    @model(
        name="model_with_depends_on",
        kind="full",
        columns={'"COL"': "int"},
        depends_on={"foo.bar"},
    )
    def my_model(context, **kwargs):
        context.resolve_table("foo")
        context.resolve_table(model_name=CONST + ".baz")

    m = model.get_registry()["model_with_depends_on"].model(
        module_path=Path("."),
        path=Path("."),
    )

    # We are not expecting the context.resolve_table() calls to be reflected in the
    # model's depends_on since we explicitly specified the depends_on argument.
    assert m.depends_on == {'"foo"."bar"'}


def test_python_model_variable_dependencies() -> None:
    @model(
        name="bla.test_model_var_dep",
        kind="full",
        columns={'"col"': "int"},
        depends_on={"@schema_name.table_name"},
    )
    def my_model(context, **kwargs):
        # Even though the argument is not statically resolvable, no error
        # is raised, because the `depends_on` property is present
        schema_name = context.var("schema_name")
        table = context.resolve_table(f"{schema_name}.table_name")

        return context.fetchdf(exp.select("*").from_(table))

    m = model.get_registry()["bla.test_model_var_dep"].model(
        module_path=Path("."),
        path=Path("."),
        variables={"schema_name": "foo"},
    )

    assert m.depends_on == {'"foo"."table_name"'}


def test_python_model_with_properties(make_snapshot):
    @model(
        name="python_model_prop",
        kind="full",
        columns={"some_col": "int"},
        session_properties={"some_string": "string_prop", "some_bool": True, "some_float": 1.0},
        physical_properties={"partition_expiration_days": 7},
        virtual_properties={"creatable_type": None},
    )
    def python_model_prop(context, **kwargs):
        context.resolve_table("foo")

    m = model.get_registry()["python_model_prop"].model(
        module_path=Path("."),
        path=Path("."),
        dialect="duckdb",
        defaults={
            "session_properties": {
                "some_string": "default_string",
                "default_value": "default_value",
            },
            "physical_properties": {
                "partition_expiration_days": 13,
                "creatable_type": "@IF(@model_kind_name != 'view', 'TRANSIENT', NULL)",
                "conditional_prop": "@IF(@model_kind_name == 'view', 'view_prop', NULL)",
            },
            "virtual_properties": {
                "creatable_type": "SECURE",
            },
        },
    )
    assert m.session_properties == {
        "some_string": "string_prop",
        "some_bool": True,
        "some_float": 1.0,
        "default_value": "default_value",
    }

    assert m.physical_properties == {
        "partition_expiration_days": exp.convert(7),
        "creatable_type": exp.maybe_parse(
            "@IF(@model_kind_name != 'view', 'TRANSIENT', NULL)", dialect="duckdb"
        ),
        "conditional_prop": exp.maybe_parse(
            "@IF(@model_kind_name == 'view', 'view_prop', NULL)", dialect="duckdb"
        ),
    }

    snapshot: Snapshot = make_snapshot(m)
    snapshot.categorize_as(SnapshotChangeCategory.BREAKING)

    # Rendering the properties will result to a TRANSIENT creatable_type and the removal of the conditional prop
    assert m.render_physical_properties(snapshots={m.fqn: snapshot}, python_env=m.python_env) == {
        "partition_expiration_days": exp.convert(7),
        "creatable_type": exp.convert("TRANSIENT"),
    }

    assert not m.virtual_properties


def test_python_models_returning_sql(assert_exp_eq) -> None:
    config = Config(model_defaults=ModelDefaultsConfig(dialect="snowflake"))
    context = Context(config=config)

    @model(
        name="model1",
        is_sql=True,
        description="A dummy model.",
        kind="full",
        dialect="snowflake",
        post_statements=["PUT file:///dir/tmp.csv @%table"],
    )
    def model1_entrypoint(evaluator: MacroEvaluator) -> exp.Select:
        return exp.select("x", "y").from_(exp.values([("1", 2), ("2", 3)], "_v", ["x", "y"]))

    @model(name="model2", is_sql=True, kind="full", dialect="snowflake")
    def model2_entrypoint(evaluator: MacroEvaluator) -> str:
        return "select * from model1"

    model1 = model.get_registry()["model1"].model(module_path=Path("."), path=Path("."))
    model2 = model.get_registry()["model2"].model(module_path=Path("."), path=Path("."))

    context.upsert_model(model1)
    context.upsert_model(model2)

    assert isinstance(model1, SqlModel)
    assert isinstance(model1.query, d.MacroFunc)
    assert model1.description == "A dummy model."
    assert model1.depends_on == set()
    assert_exp_eq(
        context.render("model1"),
        """
        SELECT
          "_V"."X" AS "X",
          "_V"."Y" AS "Y"
        FROM (VALUES ('1', 2), ('2', 3)) AS "_V"("X", "Y")
        """,
    )
    post_statement = context.get_model("model1").post_statements[0].sql(dialect="snowflake")  # type: ignore
    assert post_statement == "PUT file:///dir/tmp.csv @%table"

    assert isinstance(model2, SqlModel)
    assert isinstance(model2.query, d.MacroFunc)
    assert model2.depends_on == {'"MODEL1"'}
    assert_exp_eq(
        context.render(
            "model2",
            expand=[
                d.normalize_model_name("model1", context.default_catalog, context.config.dialect)
            ],
        ),
        """
        SELECT
          "MODEL1"."X" AS "X",
          "MODEL1"."Y" AS "Y"
        FROM (
          SELECT
            "_V"."X" AS "X",
            "_V"."Y" AS "Y"
          FROM (VALUES ('1', 2), ('2', 3)) AS "_V"("X", "Y")
        ) AS "MODEL1"
        """,
    )


def test_python_model_decorator_kind() -> None:
    # no kind specified -> default Full kind
    @model("default_kind", columns={'"COL"': "int"})
    def a_model(context):
        pass

    python_model = model.get_registry()["default_kind"].model(
        module_path=Path("."),
        path=Path("."),
    )

    assert isinstance(python_model.kind, FullKind)

    # string kind name specified
    @model("kind_string", kind="external", columns={'"COL"': "int"})
    def b_model(context):
        pass

    python_model = model.get_registry()["kind_string"].model(
        module_path=Path("."),
        path=Path("."),
    )

    assert isinstance(python_model.kind, ExternalKind)

    @model("kind_empty_dict", kind=dict(), columns={'"COL"': "int"})
    def my_model(context):
        pass

    # error if kind dict with no `name` key
    with pytest.raises(ConfigError, match="`kind` dictionary must contain a `name` key"):
        python_model = model.get_registry()["kind_empty_dict"].model(
            module_path=Path("."),
            path=Path("."),
        )

    @model("kind_dict_badname", kind=dict(name="test"), columns={'"COL"': "int"})
    def my_model_1(context):
        pass

    # error if kind dict with `name` key whose type is not a ModelKindName enum
    with pytest.raises(ConfigError, match="with a valid ModelKindName enum value"):
        python_model = model.get_registry()["kind_dict_badname"].model(
            module_path=Path("."),
            path=Path("."),
        )

    @model("kind_instance", kind=FullKind(), columns={'"COL"': "int"})
    def my_model_2(context):
        pass

    # warning if kind is ModelKind instance
    with patch.object(get_console(), "log_warning") as mock_logger:
        python_model = model.get_registry()["kind_instance"].model(
            module_path=Path("."),
            path=Path("."),
        )

        assert (
            mock_logger.call_args[0][0]
            == """Python model "kind_instance"'s `kind` argument was passed a SQLMesh `FullKind` object. This may result in unexpected behavior - provide a dictionary instead."""
        )

    # no warning with valid kind dict
    with patch.object(get_console(), "log_warning") as mock_logger:

        @model("kind_valid_dict", kind=dict(name=ModelKindName.FULL), columns={'"COL"': "int"})
        def my_model(context):
            pass

        python_model = model.get_registry()["kind_valid_dict"].model(
            module_path=Path("."),
            path=Path("."),
        )

        assert isinstance(python_model.kind, FullKind)

        assert not mock_logger.call_args


def test_python_model_decorator_col_descriptions() -> None:
    # `columns` and `column_descriptions` column names are different cases, but name normalization makes both lower
    @model("col_descriptions", columns={"col": "int"}, column_descriptions={"COL": "a column"})
    def a_model(context):
        pass

    py_model = model.get_registry()["col_descriptions"].model(
        module_path=Path("."),
        path=Path("."),
    )

    assert py_model.columns_to_types.keys() == py_model.column_descriptions.keys()

    # error: `columns` and `column_descriptions` column names are different cases, quoting preserves case
    @model(
        "col_descriptions_quoted",
        columns={'"col"': "int"},
        column_descriptions={'"COL"': "a column"},
    )
    def b_model(context):
        pass

    with patch.object(get_console(), "log_warning") as mock_logger:
        py_model = model.get_registry()["col_descriptions_quoted"].model(
            module_path=Path("."),
            path=Path("."),
        )
        assert '"COL"' not in py_model.column_descriptions
        assert (
            mock_logger.mock_calls[0].args[0]
            == "In model 'col_descriptions_quoted', a description is provided for column 'COL' but it is not a column in the model."
        )


def test_python_model_unsupported_kind() -> None:
    kinds = {
        "seed": {"name": ModelKindName.SEED, "path": "."},
        "view": {"name": ModelKindName.VIEW},
        "managed": {"name": ModelKindName.MANAGED},
        "embedded": {"name": ModelKindName.EMBEDDED},
    }

    for kindname in kinds:

        @model(f"kind_{kindname}", kind=kinds[kindname], columns={'"COL"': "int"})
        def the_kind(context):
            pass

        with pytest.raises(
            SQLMeshError, match=r".*Cannot create Python model.*doesn't support Python models"
        ):
            model.get_registry()[f"kind_{kindname}"].model(
                module_path=Path("."),
                path=Path("."),
            ).validate_definition()


def test_star_expansion(assert_exp_eq) -> None:
    context = Context(config=Config())

    model1 = load_sql_based_model(
        d.parse(
            """
        MODEL (name db.model1, kind full);

        SELECT
            id::INT AS id,
            item_id::INT AS item_id,
            ds::TEXT AS ds,
        FROM
            (VALUES
                (1, 1, '2020-01-01'),
                (1, 2, '2020-01-01'),
                (2, 1, '2020-01-01'),
                (3, 3, '2020-01-03'),
                (4, 1, '2020-01-04'),
                (5, 1, '2020-01-05'),
                (6, 1, '2020-01-06'),
                (7, 1, '2020-01-07')
            ) AS t (id, item_id, ds)
        """
        ),
        default_catalog=context.default_catalog,
    )

    model2 = load_sql_based_model(
        d.parse(
            """
        MODEL (name db.model2, kind full);

        SELECT * FROM db.model1 AS model1
        """
        ),
        default_catalog=context.default_catalog,
    )

    model3 = load_sql_based_model(
        d.parse(
            """
            MODEL(name db.model3, kind full);

            SELECT * FROM db.model2 AS model2
        """
        ),
        default_catalog=context.default_catalog,
    )

    context.upsert_model(model1)
    context.upsert_model(model2)
    context.upsert_model(model3)

    assert_exp_eq(
        context.render("db.model2", expand=["db.model1"]),
        """
        SELECT
          "model1"."id" AS "id",
          "model1"."item_id" AS "item_id",
          "model1"."ds" AS "ds"
        FROM (
          SELECT
            CAST("t"."id" AS INT) AS "id",
            CAST("t"."item_id" AS INT) AS "item_id",
            CAST("t"."ds" AS TEXT) AS "ds"
          FROM (VALUES
            (1, 1, '2020-01-01'),
            (1, 2, '2020-01-01'),
            (2, 1, '2020-01-01'),
            (3, 3, '2020-01-03'),
            (4, 1, '2020-01-04'),
            (5, 1, '2020-01-05'),
            (6, 1, '2020-01-06'),
            (7, 1, '2020-01-07')) AS "t"("id", "item_id", "ds")
        ) AS "model1"
        """,
    )
    assert_exp_eq(
        context.render("db.model3", expand=["db.model1", "db.model2"]),
        """
        SELECT
          "model2"."id" AS "id",
          "model2"."item_id" AS "item_id",
          "model2"."ds" AS "ds"
        FROM (
          SELECT
            "model1"."id" AS "id",
            "model1"."item_id" AS "item_id",
            "model1"."ds" AS "ds"
          FROM (
            SELECT
              CAST("t"."id" AS INT) AS "id",
              CAST("t"."item_id" AS INT) AS "item_id",
              CAST("t"."ds" AS TEXT) AS "ds"
            FROM (VALUES
              (1, 1, '2020-01-01'),
              (1, 2, '2020-01-01'),
              (2, 1, '2020-01-01'),
              (3, 3, '2020-01-03'),
              (4, 1, '2020-01-04'),
              (5, 1, '2020-01-05'),
              (6, 1, '2020-01-06'),
              (7, 1, '2020-01-07')) AS "t"("id", "item_id", "ds")
          ) AS "model1"
        ) AS "model2"
        """,
    )

    snapshots = context.snapshots
    snapshots['"memory"."db"."model1"'].categorize_as(SnapshotChangeCategory.BREAKING)
    snapshots['"memory"."db"."model2"'].categorize_as(SnapshotChangeCategory.BREAKING)
    snapshots['"memory"."db"."model3"'].categorize_as(SnapshotChangeCategory.BREAKING)

    assert_exp_eq(
        snapshots['"memory"."db"."model2"'].model.render_query(snapshots=snapshots),
        f"""
            SELECT
              "model1"."id" AS "id",
              "model1"."item_id" AS "item_id",
              "model1"."ds" AS "ds"
            FROM "memory"."sqlmesh__db"."db__model1__{snapshots['"memory"."db"."model1"'].version}" AS "model1"
            """,
    )

    assert_exp_eq(
        context.models['"memory"."db"."model3"'].render_query(snapshots=snapshots),
        f"""
            SELECT
              "model2"."id" AS "id",
              "model2"."item_id" AS "item_id",
              "model2"."ds" AS "ds"
            FROM "memory"."sqlmesh__db"."db__model2__{snapshots['"memory"."db"."model2"'].version}" AS "model2"
            """,
    )


def test_case_sensitivity(assert_exp_eq):
    config = Config(model_defaults=ModelDefaultsConfig(dialect="snowflake"))
    context = Context(config=config)

    source = load_sql_based_model(
        d.parse(
            """
            MODEL (name example.source, kind EMBEDDED);

            SELECT 'id' AS "id", 'name' AS "name", 'payload' AS "payload"
            """
        ),
        dialect="snowflake",
        default_catalog=context.default_catalog,
    )

    # Ensure that when manually specifying dependencies, they're normalized correctly
    downstream = load_sql_based_model(
        d.parse(
            """
            MODEL (name example.model, kind FULL, depends_on [ExAmPlE.SoUrCe]);

            SELECT JSON_EXTRACT_PATH_TEXT("payload", 'field') AS "new_field", * FROM example.source
            """
        ),
        dialect="snowflake",
        default_catalog=context.default_catalog,
    )

    context.upsert_model(source)
    context.upsert_model(downstream)

    assert_exp_eq(
        context.render("example.model"),
        """
        SELECT
          JSON_EXTRACT_PATH_TEXT("SOURCE"."payload", 'field') AS "new_field",
          "SOURCE"."id" AS "id",
          "SOURCE"."name" AS "name",
          "SOURCE"."payload" AS "payload"
        FROM (
          SELECT
            'id' AS "id",
            'name' AS "name",
            'payload' AS "payload"
        ) AS "SOURCE"
        """,
    )

    model = load_sql_based_model(
        d.parse("MODEL (name test, dialect hive); SELECT 1 AS c"),
        dialect="snowflake",
    )
    assert model.dialect == "hive"


def test_batch_size_validation():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              batch_size 100,
            ),
            columns (
              id double,
              alias varchar
            ),
            batch_size 100,
        );
    """
    )

    with pytest.raises(ConfigError):
        load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))


def test_model_cache(tmp_path: Path, mocker: MockerFixture):
    cache = ModelCache(tmp_path)

    expressions = d.parse(
        """
        MODEL (
            name db.model_sql,
        );
        SELECT 1, ds;
    """
    )

    model = load_sql_based_model([e for e in expressions if e])

    assert cache.put([model], "test_model", "test_entry_a")
    assert cache.get("test_model", "test_entry_a")[0].dict() == model.dict()

    expressions = d.parse(
        """
        MODEL (
            name db.model_seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
            ),
        );
    """
    )

    seed_model = load_sql_based_model(
        expressions, path=Path("./examples/sushi/models/test_model.sql")
    )

    assert not cache.put([seed_model], "test_model", "test_entry_b")


@pytest.mark.slow
def test_model_cache_gateway(tmp_path: Path, mocker: MockerFixture):
    init_example_project(tmp_path, engine_type="duckdb")

    db_path = str(tmp_path / "db.db")
    config = Config(
        gateways={
            "main": GatewayConfig(connection=DuckDBConnectionConfig(database=db_path)),
            "secondary": GatewayConfig(connection=DuckDBConnectionConfig()),
        },
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
    )
    Context(paths=tmp_path, config=config)

    patched_cache_put = mocker.patch("sqlmesh.utils.cache.FileCache.put")

    Context(paths=tmp_path, config=config)
    assert patched_cache_put.call_count == 0

    Context(paths=tmp_path, config=config, gateway="secondary")
    assert patched_cache_put.call_count == 2


@pytest.mark.slow
def test_model_cache_default_catalog(tmp_path: Path, mocker: MockerFixture):
    init_example_project(tmp_path, engine_type="duckdb")
    Context(paths=tmp_path)

    patched_cache_put = mocker.patch("sqlmesh.utils.cache.FileCache.put")

    Context(paths=tmp_path)
    assert patched_cache_put.call_count == 0

    with patch(
        "sqlmesh.core.engine_adapter.base.EngineAdapter.default_catalog",
        PropertyMock(return_value=None),
    ):
        Context(paths=tmp_path)
        assert patched_cache_put.call_count == 2


def test_model_ctas_query():
    expressions = d.parse(
        """
        MODEL (name `a-b-c.table`, kind FULL, dialect bigquery);
        SELECT 1 as a FROM x WHERE TRUE LIMIT 2
    """
    )

    assert (
        load_sql_based_model(expressions, dialect="bigquery").ctas_query().sql()
        == 'SELECT 1 AS "a" FROM "x" AS "x" WHERE TRUE AND FALSE LIMIT 0'
    )

    expressions = d.parse(
        """
        MODEL (name db.table, kind FULL);
        SELECT 1 as a FROM b
        """
    )

    assert (
        load_sql_based_model(expressions).ctas_query().sql()
        == 'SELECT 1 AS "a" FROM "b" AS "b" WHERE FALSE LIMIT 0'
    )

    expressions = d.parse(
        """
        MODEL (name `a-b-c.table`, kind FULL, dialect bigquery);
        SELECT 1 AS a FROM t UNION ALL SELECT 2 AS a FROM t UNION ALL SELECT 3 AS a FROM t UNION ALL SELECT 4 AS a FROM t
    """
    )

    assert (
        load_sql_based_model(expressions, dialect="bigquery").ctas_query().sql()
        == 'SELECT 1 AS "a" FROM "t" AS "t" WHERE FALSE UNION ALL SELECT 2 AS "a" FROM "t" AS "t" WHERE FALSE UNION ALL SELECT 3 AS "a" FROM "t" AS "t" WHERE FALSE UNION ALL SELECT 4 AS "a" FROM "t" AS "t" WHERE FALSE LIMIT 0'
    )

    expressions = d.parse(
        """
        MODEL (name `a-b-c.table`, kind FULL, dialect bigquery);
        SELECT 1 AS a FROM t UNION ALL SELECT 2 AS a FROM t
    """
    )

    assert (
        load_sql_based_model(expressions, dialect="bigquery").ctas_query().sql()
        == 'SELECT 1 AS "a" FROM "t" AS "t" WHERE FALSE UNION ALL SELECT 2 AS "a" FROM "t" AS "t" WHERE FALSE LIMIT 0'
    )

    expressions = d.parse(
        """
        MODEL (name `a-b-c.table`, kind FULL, dialect bigquery);
        SELECT 1 AS a FROM t UNION ALL SELECT 2 AS a FROM t ORDER BY 1
    """
    )

    assert (
        load_sql_based_model(expressions, dialect="bigquery").ctas_query().sql()
        == 'SELECT 1 AS "a" FROM "t" AS "t" WHERE FALSE UNION ALL SELECT 2 AS "a" FROM "t" AS "t" WHERE FALSE ORDER BY 1 LIMIT 0'
    )

    expressions = d.parse(
        """
        MODEL (name `a-b-c.table`, kind FULL, dialect bigquery);
        WITH RECURSIVE a AS (
            SELECT * FROM x
        ), b AS (
            SELECT * FROM a UNION ALL SELECT * FROM a
        )
        SELECT * FROM b

    """
    )

    assert (
        load_sql_based_model(expressions, dialect="bigquery").ctas_query().sql()
        == 'WITH RECURSIVE "a" AS (SELECT * FROM "x" AS "x" WHERE FALSE), "b" AS (SELECT * FROM "a" AS "a" WHERE FALSE UNION ALL SELECT * FROM "a" AS "a" WHERE FALSE) SELECT * FROM "b" AS "b" WHERE FALSE LIMIT 0'
    )

    expressions = d.parse(
        """
        MODEL (name `a-b-c.table`, kind FULL, dialect bigquery);
        WITH RECURSIVE a AS (
            SELECT * FROM (SELECT * FROM (SELECT * FROM x))
        ), b AS (
            SELECT * FROM a UNION ALL SELECT * FROM a
        )
        SELECT * FROM b

    """
    )

    assert (
        load_sql_based_model(expressions, dialect="bigquery").ctas_query().sql()
        == 'WITH RECURSIVE "a" AS (SELECT * FROM (SELECT * FROM (SELECT * FROM "x" AS "x" WHERE FALSE) AS "_q_0" WHERE FALSE) AS "_q_1" WHERE FALSE), "b" AS (SELECT * FROM "a" AS "a" WHERE FALSE UNION ALL SELECT * FROM "a" AS "a" WHERE FALSE) SELECT * FROM "b" AS "b" WHERE FALSE LIMIT 0'
    )

    expressions = d.parse(
        """
        MODEL (name `a-b-c.table`, kind FULL, dialect bigquery);
        WITH RECURSIVE a AS (
            WITH nested_a AS (
                SELECT * FROM (SELECT * FROM (SELECT * FROM x))
            )
            SELECT * FROM nested_a
        ), b AS (
            SELECT * FROM a UNION ALL SELECT * FROM a
        )
        SELECT * FROM b

    """
    )

    assert (
        load_sql_based_model(expressions, dialect="bigquery").ctas_query().sql()
        == 'WITH RECURSIVE "a" AS (WITH "nested_a" AS (SELECT * FROM (SELECT * FROM (SELECT * FROM "x" AS "x" WHERE FALSE) AS "_q_0" WHERE FALSE) AS "_q_1" WHERE FALSE) SELECT * FROM "nested_a" AS "nested_a" WHERE FALSE), "b" AS (SELECT * FROM "a" AS "a" WHERE FALSE UNION ALL SELECT * FROM "a" AS "a" WHERE FALSE) SELECT * FROM "b" AS "b" WHERE FALSE LIMIT 0'
    )


def test_is_breaking_change():
    model = create_external_model("a", columns={"a": "int", "limit": "int"})
    assert model.is_breaking_change(create_external_model("a", columns={"a": "int"})) is False
    assert model.is_breaking_change(create_external_model("a", columns={"a": "text"})) is None
    assert (
        model.is_breaking_change(create_external_model("a", columns={"a": "int", "limit": "int"}))
        is False
    )
    assert (
        model.is_breaking_change(
            create_external_model("a", columns={"a": "int", "limit": "int", "c": "int"})
        )
        is None
    )


def test_parse_expression_list_with_jinja():
    input = [
        "JINJA_STATEMENT_BEGIN;\n{{ log('log message') }}\nJINJA_END;",
        "GRANT SELECT ON TABLE foo TO DEV",
    ]
    assert input == [val.sql() for val in parse_expression(SqlModel, input, None)]


def test_no_depends_on_runtime_jinja_query():
    @macro()
    def runtime_macro(evaluator, **kwargs) -> None:
        from sqlmesh.utils.errors import ParsetimeAdapterCallError

        raise ParsetimeAdapterCallError("")

    expressions = d.parse(
        """
        MODEL (name db.table);

        JINJA_QUERY_BEGIN;
        SELECT {{ runtime_macro() }} as a FROM b
        JINJA_QUERY_END;
        """
    )

    model = load_sql_based_model(expressions)
    with pytest.raises(
        ConfigError,
        match=r"Dependencies must be provided explicitly for models that can be rendered only at runtime",
    ):
        model.validate_definition()


@use_terminal_console
def test_update_schema(tmp_path: Path):
    expressions = d.parse(
        """
        MODEL (name db.table);

        SELECT * FROM table_a JOIN table_b
        """
    )

    model = load_sql_based_model(expressions)
    schema = MappingSchema(normalize=False)

    # Even though the partial schema is applied, we won't optimize the model
    schema.add_table('"table_a"', {"a": exp.DataType.build("int")})
    model.update_schema(schema)
    assert model.mapping_schema == {'"table_a"': {"a": "INT"}}

    ctx = Context(
        config=Config(linter=LinterConfig(enabled=True, warn_rules=["ALL"])), paths=tmp_path
    )
    with patch.object(get_console(), "log_warning") as mock_logger:
        ctx.upsert_model(model)
        ctx.plan_builder("dev")
        assert (
            missing_schema_warning_msg('"db"."table"', ('"table_b"',))
            in mock_logger.call_args[0][0]
        )

    schema.add_table('"table_b"', {"b": exp.DataType.build("int")})
    model.update_schema(schema)
    assert model.mapping_schema == {
        '"table_a"': {"a": "INT"},
        '"table_b"': {"b": "INT"},
    }
    model.render_query(needs_optimization=True)


@use_terminal_console
def test_missing_schema_warnings(tmp_path: Path):
    full_schema = MappingSchema(
        {
            "a": {"x": exp.DataType.build("int")},
            "b": {"y": exp.DataType.build("int")},
        },
        normalize=False,
    )

    partial_schema = MappingSchema(
        {
            '"a"': {"x": exp.DataType.build("int")},
        },
    )

    console = get_console()

    ctx = Context(
        config=Config(linter=LinterConfig(enabled=True, warn_rules=["ALL"])), paths=tmp_path
    )

    # star, no schema, no deps
    with patch.object(console, "log_warning") as mock_logger:
        model = load_sql_based_model(d.parse("MODEL (name test); SELECT * FROM (SELECT 1 a) x"))
        model.render_query(needs_optimization=True)
        mock_logger.assert_not_called()

    # star, full schema
    with patch.object(console, "log_warning") as mock_logger:
        model = load_sql_based_model(d.parse("MODEL (name test); SELECT * FROM a CROSS JOIN b"))
        model.update_schema(full_schema)
        model.render_query(needs_optimization=True)
        mock_logger.assert_not_called()

    # star, partial schema
    with patch.object(console, "log_warning") as mock_logger:
        model = load_sql_based_model(d.parse("MODEL (name test); SELECT * FROM a CROSS JOIN b"))
        model.update_schema(partial_schema)
        ctx.upsert_model(model)
        ctx.plan_builder("dev")
        assert missing_schema_warning_msg('"test"', ('"b"',)) in mock_logger.call_args[0][0]

    # star, no schema
    with patch.object(console, "log_warning") as mock_logger:
        model = load_sql_based_model(d.parse("MODEL (name test); SELECT * FROM b JOIN a"))
        ctx.upsert_model(model)
        ctx.plan_builder("dev")
        assert missing_schema_warning_msg('"test"', ('"a"', '"b"')) in mock_logger.call_args[0][0]

    # no star, full schema
    with patch.object(console, "log_warning") as mock_logger:
        model = load_sql_based_model(
            d.parse("MODEL (name test); SELECT x::INT FROM a CROSS JOIN b")
        )
        model.update_schema(full_schema)
        model.render_query(needs_optimization=True)
        mock_logger.assert_not_called()

    # no star, partial schema
    with patch.object(console, "log_warning") as mock_logger:
        model = load_sql_based_model(
            d.parse("MODEL (name test); SELECT x::INT FROM a CROSS JOIN b")
        )
        model.update_schema(partial_schema)
        model.render_query(needs_optimization=True)
        mock_logger.assert_not_called()

    # no star, empty schema
    with patch.object(console, "log_warning") as mock_logger:
        model = load_sql_based_model(
            d.parse("MODEL (name test); SELECT x::INT FROM a CROSS JOIN b")
        )
        model.render_query(needs_optimization=True)
        mock_logger.assert_not_called()


def test_user_provided_depends_on():
    for l_delim, r_delim in (("(", ")"), ("[", "]")):
        expressions = d.parse(
            f"""
            MODEL (name db.table, depends_on {l_delim}table_b{r_delim});

            SELECT a FROM table_a
            """
        )

        model = load_sql_based_model(expressions)
        assert model.depends_on == {'"table_a"', '"table_b"'}, f"Delimiters {l_delim}, {r_delim}"


def test_check_schema_mapping_when_rendering_at_runtime(assert_exp_eq):
    expressions = d.parse(
        """
        MODEL (name db.table, depends_on [table_b]);

        SELECT * FROM table_a JOIN table_b
        """
    )

    model = load_sql_based_model(expressions)

    # Simulate a query that cannot be rendered at parse time.
    with patch.object(SqlModel, "render_query", return_value=None) as render_query_mock:
        assert not model.columns_to_types

        schema = MappingSchema(normalize=False)
        schema.add_table('"table_b"', {"b": exp.DataType.build("int")})
        model.update_schema(schema)

    assert '"table_b"' in model.mapping_schema
    assert model.depends_on == {'"table_b"'}

    render_query_mock.assert_called_once()

    # Simulate rendering at runtime.
    assert_exp_eq(
        model.render_query(), """SELECT * FROM "table_a" AS "table_a", "table_b" AS "table_b" """
    )


def test_model_normalization():
    expr = d.parse(
        """
        MODEL (
            name `project-1.db.tbl`,
            kind FULL,
            columns ( a STRUCT <`a` INT64> ),
            partitioned_by foo(`ds`),
            dialect bigquery,
            grain [id, ds]
        );
        SELECT * FROM project-1.db.raw
        """
    )

    model = SqlModel.parse_raw(load_sql_based_model(expr, depends_on={"project-2.db.raw"}).json())
    assert model.name == "`project-1.db.tbl`"
    assert model.columns_to_types["a"].sql(dialect="bigquery") == "STRUCT<`a` INT64>"
    assert model.partitioned_by[0].sql(dialect="bigquery") == "foo(`ds`)"
    # The values are quoted with double quotes instead of backticks because of how model name normalization works
    # since it normalizes values according to the target dialect but not the quotes
    assert model.depends_on == {'"project-1"."db"."raw"', '"project-2"."db"."raw"'}

    expr = d.parse(
        """
        MODEL (
            name foo,
            kind INCREMENTAL_BY_TIME_RANGE (
              time_column a
            ),
            columns (a int),
            partitioned_by foo("ds"),
            dialect snowflake,
            grain [id, ds],
            tags (pii, fact),
            clustered_by a
        );
        SELECT * FROM bla
        """
    )

    model = SqlModel.parse_raw(load_sql_based_model(expr).json())
    assert model.name == "foo"
    assert model.time_column.column == exp.to_column("A", quoted=True)
    assert model.columns_to_types["A"].sql(dialect="snowflake") == "INT"
    assert model.partitioned_by[0].sql(dialect="snowflake") == '"A"'
    assert model.partitioned_by[1].sql(dialect="snowflake") == 'FOO("ds")'
    assert model.tags == ["pii", "fact"]
    assert model.clustered_by == [exp.to_column('"A"')]
    assert model.depends_on == {'"BLA"'}

    # Check possible variations of unique_key definitions
    for key in ("""[a, COALESCE(b, ''), "c"]""", """(a, COALESCE(b, ''), "c")"""):
        expr = d.parse(
            f"""
            MODEL (
                name foo,
                kind INCREMENTAL_BY_UNIQUE_KEY (
                    unique_key {key}
                ),
                dialect snowflake
            );

            SELECT
              x.a AS a,
              x.b AS b,
              x."c" AS c
            FROM test.x AS x
            """
        )
        model = SqlModel.parse_raw(load_sql_based_model(expr).json())
        assert model.unique_key == [
            exp.column("A", quoted=True),
            exp.func("COALESCE", exp.column("B", quoted=True), "''"),
            exp.column("c", quoted=True),
        ]

    expr = d.parse(
        """
        MODEL (
            name foo,
            dialect snowflake,
            kind INCREMENTAL_BY_UNIQUE_KEY(unique_key a),
              audits (
                not_null(COLUMNS := id)
            ),
        );

        SELECT x.a AS a FROM test.x AS x
        """
    )
    model = SqlModel.parse_raw(load_sql_based_model(expr).json())
    assert model.unique_key == [exp.column("A", quoted=True)]
    # we should never normalize the model meta, additionally, we should force lower case
    assert model.audits[0][1] == {"columns": exp.column("id")}

    model = create_sql_model(
        "foo",
        parse_one("SELECT * FROM bla"),
        columns={"a": "int"},
        kind=IncrementalByTimeRangeKind(time_column=exp.column("a"), dialect="snowflake"),
        dialect="snowflake",
        grain=[exp.to_column("id"), exp.to_column("ds")],
        tags=["pii", "fact"],
        clustered_by=exp.to_column("a"),
    )
    assert model.name == "foo"
    assert model.time_column.column == exp.column("A", quoted=True)
    assert model.columns_to_types["A"].sql(dialect="snowflake") == "INT"
    assert model.tags == ["pii", "fact"]
    assert model.clustered_by == [exp.to_column('"A"')]
    assert model.depends_on == {'"BLA"'}

    model = create_sql_model(
        "foo",
        parse_one("SELECT * FROM bla"),
        kind=IncrementalByTimeRangeKind(time_column=exp.to_column("a")),
        dialect="snowflake",
        grain=[exp.to_column("id"), exp.to_column("ds")],
        tags=["pii", "fact"],
        clustered_by=[exp.column("a"), exp.column("b")],
    )
    assert model.clustered_by == [exp.to_column('"A"'), exp.to_column('"B"')]

    model = create_sql_model(
        "foo",
        parse_one("SELECT * FROM bla"),
        kind=IncrementalByTimeRangeKind(time_column=exp.to_column("a")),
        dialect="snowflake",
        grain=[exp.to_column("id"), exp.to_column("ds")],
        tags=["pii", "fact"],
        clustered_by=["a", "b"],
    )
    assert model.clustered_by == [exp.to_column('"A"'), exp.to_column('"B"')]


def test_incremental_unmanaged_validation():
    model = create_sql_model(
        "a",
        d.parse_one("SELECT a, ds FROM table_a"),
        kind=IncrementalUnmanagedKind(insert_overwrite=True),
    )

    with pytest.raises(
        ConfigError,
        match=r"Unmanaged incremental models with insert / overwrite enabled must specify the partitioned_by field.*",
    ):
        model.validate_definition()

    model = model.copy(update={"partitioned_by_": [exp.to_column("ds")]})
    model.validate_definition()


def test_incremental_unmanaged():
    expr = d.parse(
        """
        MODEL (
            name foo,
            kind INCREMENTAL_UNMANAGED
        );

        SELECT x.a AS a FROM test.x AS x
        """
    )

    model = load_sql_based_model(expressions=expr)

    assert isinstance(model.kind, IncrementalUnmanagedKind)
    assert not model.kind.insert_overwrite

    expr = d.parse(
        """
        MODEL (
            name foo,
            kind INCREMENTAL_UNMANAGED (
                insert_overwrite true
            ),
            partitioned_by a
        );

        SELECT x.a AS a FROM test.x AS x
        """
    )

    model = load_sql_based_model(expressions=expr)
    assert isinstance(model.kind, IncrementalUnmanagedKind)
    assert model.kind.insert_overwrite


def test_custom_interval_unit():
    assert (
        load_sql_based_model(
            d.parse("MODEL (name db.table, interval_unit FIVE_MINUTE); SELECT a FROM tbl;")
        ).interval_unit
        == IntervalUnit.FIVE_MINUTE
    )

    assert (
        load_sql_based_model(
            d.parse("MODEL (name db.table, interval_unit hour); SELECT a FROM tbl;")
        ).interval_unit
        == IntervalUnit.HOUR
    )

    assert (
        load_sql_based_model(
            d.parse("MODEL (name db.table, interval_unit Hour, cron '@daily'); SELECT a FROM tbl;")
        ).interval_unit
        == IntervalUnit.HOUR
    )

    assert (
        load_sql_based_model(
            d.parse("MODEL (name db.table, cron '@daily'); SELECT a FROM tbl;")
        ).interval_unit
        == IntervalUnit.DAY
    )

    assert (
        load_sql_based_model(
            d.parse("MODEL (name db.table, cron '0 5 * * *'); SELECT a FROM tbl;")
        ).interval_unit
        == IntervalUnit.DAY
    )

    assert (
        load_sql_based_model(
            d.parse(
                "MODEL (name db.table, cron '0 5 * * *', interval_unit 'quarter_hour'); SELECT a FROM tbl;"
            )
        ).interval_unit
        == IntervalUnit.QUARTER_HOUR
    )

    with pytest.raises(
        ConfigError, match=r"Cron '@daily' cannot be more frequent than interval unit 'month'."
    ):
        load_sql_based_model(
            d.parse("MODEL (name db.table, interval_unit month); SELECT a FROM tbl;")
        )

    with pytest.raises(
        ConfigError,
        match=r"Cron '@hourly' cannot be more frequent than interval unit 'day'. If this is intentional, set allow_partials to True.",
    ):
        load_sql_based_model(
            d.parse("MODEL (name db.table, interval_unit Day, cron '@hourly'); SELECT a FROM tbl;")
        )


def test_interval_unit_larger_than_cron_period():
    # The interval unit can be larger than the cron period if allow_partials is True
    model = load_sql_based_model(
        d.parse(
            "MODEL (name db.table, interval_unit day, cron '@hourly', allow_partials TRUE); SELECT a FROM tbl;"
        )
    )
    assert model.interval_unit == IntervalUnit.DAY
    assert model.cron == "@hourly"
    assert model.allow_partials

    with pytest.raises(
        ConfigError,
        match=r"Cron '@hourly' cannot be more frequent than interval unit 'day'. If this is intentional, set allow_partials to True.",
    ):
        load_sql_based_model(
            d.parse("MODEL (name db.table, interval_unit day, cron '@hourly'); SELECT a FROM tbl;")
        )

    with pytest.raises(
        ConfigError,
        match=r"Cron '@hourly' cannot be more frequent than interval unit 'day'. If this is intentional, set allow_partials to True.",
    ):
        load_sql_based_model(
            d.parse(
                "MODEL (name db.table, interval_unit day, cron '@hourly', allow_partials FALSE); SELECT a FROM tbl;"
            )
        )


def test_model_physical_properties() -> None:
    # Validate python model table properties
    @model(
        "my_model",
        kind="full",
        columns={"id": "int"},
        physical_properties={
            "format": "PARQUET",
            "bucket_count": 0,
            "orc_bloom_filter_fpp": 0.05,
            "auto_purge": False,
        },
    )
    def my_model(context, **kwargs):
        pass

    python_model = model.get_registry()["my_model"].model(
        module_path=Path("."),
        path=Path("."),
    )

    assert python_model.physical_properties == {
        "format": exp.Literal.string("PARQUET"),
        "bucket_count": exp.Literal.number(0),
        "orc_bloom_filter_fpp": exp.Literal.number(0.05),
        "auto_purge": exp.false(),
    }

    # Validate a tuple.
    sql_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            physical_properties (
                key_a = 'value_a',
                'key_b' = 1,
                key_c = true,
                "key_d" = 2.0,
            )
        );
        SELECT a FROM tbl;
        """
        )
    )
    assert sql_model.physical_properties == {
        "key_a": exp.convert("value_a"),
        "key_b": exp.convert(1),
        "key_c": exp.convert(True),
        "key_d": exp.convert(2.0),
    }
    assert sql_model.physical_properties_ == d.parse_one(
        """(key_a = 'value_a', 'key_b' = 1, key_c = TRUE, "key_d" = 2.0)"""
    )

    # Validate a tuple with one item.
    sql_model = load_sql_based_model(
        d.parse(
            """
            MODEL (
                name test_schema.test_model,
                physical_properties (key_a = 'value_a')
            );
            SELECT a FROM tbl;
            """
        )
    )
    assert sql_model.physical_properties == {"key_a": exp.convert("value_a")}
    assert (
        sql_model.physical_properties_.sql()  # type: ignore
        == exp.Tuple(expressions=[d.parse_one("key_a = 'value_a'")]).sql()
    )

    # Validate an array.
    sql_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            physical_properties [
                key_a = 'value_a',
                'key_b' = 1,
            ]
        );
        SELECT a FROM tbl;
        """
        )
    )
    assert sql_model.physical_properties == {
        "key_a": exp.convert("value_a"),
        "key_b": exp.convert(1),
    }
    assert sql_model.physical_properties_ == d.parse_one("""(key_a = 'value_a', 'key_b' = 1)""")

    # Validate empty.
    sql_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model
        );
        SELECT a FROM tbl;
        """
        )
    )
    assert sql_model.physical_properties == {}
    assert sql_model.physical_properties_ is None

    # Validate sql expression.
    sql_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            physical_properties [
                key = ['value']
            ]
        );
        SELECT a FROM tbl;
        """
        )
    )
    assert sql_model.physical_properties == {"key": d.parse_one("['value']")}
    assert sql_model.physical_properties_ == exp.Tuple(expressions=[d.parse_one("key = ['value']")])

    # Validate dict parsing.
    sql_model = create_sql_model(
        name="test_schema.test_model",
        query=d.parse_one("SELECT a FROM tbl"),
        physical_properties={
            "key_a": exp.Literal.string("value_a"),
            "key_b": exp.Literal.number(1),
            "key_c": exp.true(),
            "key_d": exp.Literal.number(2.0),
        },
    )
    assert sql_model.physical_properties == {
        "key_a": exp.convert("value_a"),
        "key_b": exp.convert(1),
        "key_c": exp.convert(True),
        "key_d": exp.convert(2.0),
    }
    assert sql_model.physical_properties_ == d.parse_one(
        """('key_a' = 'value_a', 'key_b' = 1, 'key_c' = TRUE, 'key_d' = 2.0)"""
    )

    with pytest.raises(
        ConfigError,
        match=r"Invalid property 'invalid'. Properties must be specified as key-value pairs <key> = <value>. ",
    ):
        load_sql_based_model(
            d.parse(
                """
                MODEL (
                    name test_schema.test_model,
                    physical_properties [
                        invalid
                    ]
                );
                SELECT a FROM tbl;
                """
            )
        )


def test_model_physical_properties_labels() -> None:
    sql_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            physical_properties [
                labels = [('test-label', 'label-value')]
            ]
        );
        SELECT a FROM tbl;
        """
        )
    )
    assert sql_model.physical_properties == {"labels": exp.array("('test-label', 'label-value')")}


def test_physical_and_virtual_table_properties() -> None:
    sql_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            physical_properties (
                partition_expiration_days = 7,
                labels = [('test-physical-label', 'label-physical-value')],
            ),
            virtual_properties (
                labels = [('test-virtual-label', 'label-virtual-value')],
            )
        );
        SELECT a FROM tbl;
        """
        )
    )
    assert sql_model.physical_properties == {
        "partition_expiration_days": exp.convert(7),
        "labels": exp.array("('test-physical-label', 'label-physical-value')"),
    }

    assert sql_model.virtual_properties == {
        "labels": exp.array("('test-virtual-label', 'label-virtual-value')"),
    }


def test_model_table_properties() -> None:
    # Ensure backward compatibility to table_properties.
    sql_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            table_properties (
                key_a = 'value_a',
                'key_b' = 1,
                key_c = true,
                "key_d" = 2.0,
            )
        );
        SELECT a FROM tbl;
        """
        )
    )
    assert sql_model.physical_properties == {
        "key_a": exp.convert("value_a"),
        "key_b": exp.convert(1),
        "key_c": exp.convert(True),
        "key_d": exp.convert(2.0),
    }
    assert sql_model.physical_properties_ == d.parse_one(
        """(key_a = 'value_a', 'key_b' = 1, key_c = TRUE, "key_d" = 2.0)"""
    )

    sql_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            table_properties (
                partition_expiration_days = 7,
            )
        );
        SELECT a FROM tbl;
        """
        )
    )
    assert sql_model.physical_properties == {
        "partition_expiration_days": exp.convert(7),
    }
    assert sql_model.physical_properties_ == d.parse_one("""(partition_expiration_days = 7,)""")


def test_model_table_properties_conflicts() -> None:
    # Throw an error on conflicting usage of table_properties and physical_properties.
    with pytest.raises(ConfigError, match=r"Cannot use argument 'table_properties'*"):
        sql_model = load_sql_based_model(
            d.parse(
                """
                MODEL (
                    name test_schema.test_model,
                    table_properties (
                        key_a = 'value_a',
                        'key_b' = 1,
                        key_c = true,
                        "key_d" = 2.0,
                    ),
                    physical_properties (key_a = 'value_a')
                );
                SELECT a FROM tbl;
                """
            )
        )
        sql_model.physical_properties


def test_session_properties_on_model_and_project(sushi_context):
    model_defaults = ModelDefaultsConfig(
        session_properties={
            "some_bool": False,
            "quoted_identifier": "value_you_wont_see",
            "project_level_property": "project_property",
        },
        physical_properties={
            "warehouse": "small",
            "target_lag": "10 minutes",
        },
        virtual_properties={"creatable_type": "SECURE"},
    )

    model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            session_properties (
                'spark.executor.cores' = 2,
                'spark.executor.memory' = '1G',
                some_bool = True,
                some_float = 0.1,
                quoted_identifier = "quoted identifier",
                unquoted_identifier = unquoted_identifier,
            ),
            physical_properties (
                target_lag = '1 hour'
            ),
        );
        SELECT a FROM tbl;
        """,
            default_dialect="snowflake",
        ),
        defaults=model_defaults.dict(),
    )

    assert model.session_properties == {
        "spark.executor.cores": 2,
        "spark.executor.memory": "1G",
        "some_bool": True,
        "some_float": 0.1,
        "quoted_identifier": exp.column("quoted identifier", quoted=True),
        "unquoted_identifier": exp.column("unquoted_identifier", quoted=False),
        "project_level_property": "project_property",
    }

    assert model.physical_properties == {
        "warehouse": exp.convert("small"),
        "target_lag": exp.convert("1 hour"),
    }

    assert model.virtual_properties == {
        "creatable_type": exp.convert("SECURE"),
    }


def test_project_level_properties(sushi_context):
    model_defaults = ModelDefaultsConfig(
        session_properties={
            "some_bool": False,
            "some_float": 0.1,
            "project_level_property": "project_property",
        },
        physical_properties={
            "warehouse": "small",
            "target_lag": "1 hour",
        },
        virtual_properties={"creatable_type": "SECURE"},
        enabled=False,
        allow_partials=True,
        interval_unit="quarter_hour",
        optimize_query=True,
        cron="@hourly",
    )

    model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            kind FULL,
            virtual_properties (
                creatable_type = None
            ),
        );
        SELECT a FROM tbl;
        """,
            default_dialect="snowflake",
        ),
        defaults=model_defaults.dict(),
    )

    # Validate use of project wide defaults
    assert not model.enabled
    assert model.allow_partials
    assert model.interval_unit == IntervalUnit.QUARTER_HOUR
    assert model.optimize_query
    assert model.cron == "@hourly"

    assert model.session_properties == {
        "some_bool": False,
        "some_float": 0.1,
        "project_level_property": "project_property",
    }

    assert model.physical_properties == {
        "warehouse": exp.convert("small"),
        "target_lag": exp.convert("1 hour"),
    }

    # Validate disabling global property
    assert not model.virtual_properties

    model_2 = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model_2,
            kind FULL,
            allow_partials False,
            interval_unit hour,
            cron '@daily'

        );
        SELECT a, b FROM tbl;
        """,
            default_dialect="snowflake",
        ),
        defaults=model_defaults.dict(),
    )

    # Validate overriding of project wide defaults
    assert not model_2.allow_partials
    assert model_2.interval_unit == IntervalUnit.HOUR
    assert model_2.cron == "@daily"


def test_conditional_physical_properties(make_snapshot):
    model_defaults = ModelDefaultsConfig(
        physical_properties={
            "creatable_type": "@IF(@model_kind_name != 'VIEW', 'TRANSIENT', NULL)"
        },
    )

    full_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_full_model,
            kind FULL,
        );
        SELECT a FROM tbl;
        """,
            default_dialect="snowflake",
        ),
        defaults=model_defaults.dict(),
    )

    view_model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_view_model_kind,
            kind VIEW,
        );
        SELECT a FROM tbl;
        """,
            default_dialect="snowflake",
        ),
        defaults=model_defaults.dict(),
    )

    # load time is a no-op
    assert (
        view_model.physical_properties
        == full_model.physical_properties
        == {
            "creatable_type": exp.maybe_parse(
                "@IF(@model_kind_name != 'VIEW', 'TRANSIENT', NULL)", dialect="snowflake"
            )
        }
    )

    # substitution occurs at runtime
    snapshot: Snapshot = make_snapshot(full_model)
    snapshot_view: Snapshot = make_snapshot(view_model)
    snapshot.categorize_as(SnapshotChangeCategory.BREAKING)

    # Validate use of TRANSIENT type for FULL model
    assert full_model.render_physical_properties(
        snapshots={full_model.fqn: snapshot, view_model.fqn: snapshot_view}
    ) == {"creatable_type": exp.Literal(this="TRANSIENT", is_string=True)}

    # Validate disabling the creatable_type property for VIEW model
    assert (
        view_model.render_physical_properties(
            snapshots={full_model.fqn: snapshot, view_model.fqn: snapshot_view}
        )
        == {}
    )


def test_project_level_properties_python_model():
    model_defaults = {
        "physical_properties": {
            "partition_expiration_days": 13,
            "creatable_type": "TRANSIENT",
        },
        "description": "general model description",
        "enabled": False,
        "allow_partials": True,
        "interval_unit": "quarter_hour",
        "optimize_query": True,
    }

    @model(
        name="python_model_t_defaults",
        kind="full",
        columns={"some_col": "int"},
        physical_properties={"partition_expiration_days": 7},
    )
    def python_model_prop(context, **kwargs):
        context.resolve_table("foo")

    m = model.get_registry()["python_model_t_defaults"].model(
        module_path=Path("."),
        path=Path("."),
        dialect="duckdb",
        defaults=model_defaults,
    )

    assert m.physical_properties == {
        "partition_expiration_days": exp.convert(7),
        "creatable_type": exp.convert("TRANSIENT"),
    }

    # Even if in the project wide defaults these are ignored for python models
    assert not m.optimize_query

    assert not m.enabled
    assert m.allow_partials
    assert m.interval_unit == IntervalUnit.QUARTER_HOUR


def test_model_defaults_macros(make_snapshot):
    model_defaults = ModelDefaultsConfig(
        table_format="@IF(@gateway = 'dev', 'iceberg', NULL)",
        cron="@cron_macro",
        storage_format="@IF(@gateway = 'local', 'parquet', NULL)",
        optimize_query="@IF(@gateway = 'dev', True, False)",
        enabled="@IF(@gateway = 'dev', True, False)",
        allow_partials="@IF(@gateway = 'local', True, False)",
        interval_unit="@IF(@gateway = 'local', 'quarter_hour', 'day')",
        start="@IF(@gateway = 'dev', '1 month ago', '2024-01-01')",
        session_properties={
            "spark.executor.cores": "@IF(@gateway = 'dpev', 1, 2)",
            "spark.executor.memory": "1G",
            "unset_property": "@IF(@model_kind_name = 'FULL', 'NOTSET', NULL)",
        },
        physical_properties={
            "partition_expiration_days": 13,
            "creatable_type": "@IF(@model_kind_name = 'FULL', 'TRANSIENT', NULL)",
        },
        virtual_properties={
            "creatable_type": "@create_type",
            "unset_property": "@IF(@model_kind_name = 'FULL', 'NOTSET', NULL)",
        },
    )

    model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            physical_properties (
                target_lag = '1 hour'
            ),
        );
        SELECT a FROM tbl;
        """,
            default_dialect="snowflake",
        ),
        defaults=model_defaults.dict(),
        variables={"gateway": "dev", "create_type": "SECURE", "cron_macro": "@daily"},
    )

    snapshot: Snapshot = make_snapshot(model)
    snapshot.categorize_as(SnapshotChangeCategory.BREAKING)

    # Validate rendering of model defaults
    assert model.optimize_query
    assert model.enabled
    assert model.start == "1 month ago"
    assert not model.allow_partials
    assert model.interval_unit == IntervalUnit.DAY
    assert model.table_format == "iceberg"
    assert model.cron == "@daily"

    # Validate disabling of conditional model default
    assert not model.storage_format

    # The model defaults properties won't be rendered at load time
    assert model.session_properties == {
        "spark.executor.cores": exp.maybe_parse(
            "@IF(@gateway = 'dpev', 1, 2)", dialect="snowflake"
        ),
        "spark.executor.memory": "1G",
        "unset_property": exp.maybe_parse(
            "@IF(@model_kind_name = 'FULL', 'NOTSET', NULL)", dialect="snowflake"
        ),
    }

    assert model.physical_properties == {
        "partition_expiration_days": exp.convert(13),
        "creatable_type": exp.maybe_parse(
            "@IF(@model_kind_name = 'FULL', 'TRANSIENT', NULL)", dialect="snowflake"
        ),
        "target_lag": exp.convert("1 hour"),
    }

    assert model.virtual_properties == {
        "creatable_type": d.MacroVar(this="create_type"),
        "unset_property": exp.maybe_parse(
            "@IF(@model_kind_name = 'FULL', 'NOTSET', NULL)", dialect="snowflake"
        ),
    }

    # Validate the correct rendering and removal of conditional properties
    assert model.render_session_properties(snapshots={model.fqn: snapshot}) == {
        "spark.executor.cores": exp.convert(2),
        "spark.executor.memory": "1G",
    }

    assert model.render_physical_properties(snapshots={model.fqn: snapshot}) == {
        "partition_expiration_days": exp.convert(13),
        "target_lag": exp.convert("1 hour"),
    }

    assert model.render_virtual_properties(snapshots={model.fqn: snapshot}) == {
        "creatable_type": exp.convert("SECURE"),
    }


def test_model_defaults_macros_python_model(make_snapshot):
    model_defaults = {
        "physical_properties": {
            "partition_expiration_days": 13,
            "creatable_type": "@IF(@model_kind_name = 'FULL', 'TRANSIENT', NULL)",
        },
        "cron": "@cron_macro_expr",
        "table_format": "@IF(@gateway = 'local', 'iceberg', NULL)",
        "storage_format": "@IF(@gateway = 'dev', 'parquet', NULL)",
        "optimize_query": "@IF(@gateway = 'local', True, False)",
        "enabled": "@IF(@gateway = 'local', True, False)",
        "allow_partials": "@IF(@gateway = 'local', True, False)",
        "interval_unit": "@IF(@gateway = 'local', 'quarter_hour', 'day')",
        "start": "@IF(@gateway = 'dev', '1 month ago', '2024-01-01')",
        "virtual_properties": {"creatable_type": "@create_type"},
        "session_properties": {
            "spark.executor.cores": "@IF(@gateway = 'dev', 1, 2)",
            "spark.executor.memory": "1G",
        },
    }

    @model(
        name="python_model_defaults_macro",
        kind="full",
        columns={"some_col": "int"},
        physical_properties={"partition_expiration_days": 7},
    )
    def python_model_prop_macro(context, **kwargs):
        context.resolve_table("foo")

    m = model.get_registry()["python_model_defaults_macro"].model(
        module_path=Path("."),
        path=Path("."),
        dialect="duckdb",
        defaults=model_defaults,
        variables={"gateway": "local", "create_type": "SECURE", "cron_macro_expr": "0 */2 * * *"},
    )

    # Even if in the project wide defaults this is ignored for python models
    assert not m.optimize_query

    # Validate rendering of model defaults
    assert m.cron == "0 */2 * * *"
    assert m.enabled
    assert m.start == "2024-01-01"
    assert m.allow_partials
    assert m.interval_unit == IntervalUnit.QUARTER_HOUR
    assert m.table_format == "iceberg"

    # Validate disabling attribute dynamically
    assert not m.storage_format

    snapshot: Snapshot = make_snapshot(m)
    snapshot.categorize_as(SnapshotChangeCategory.BREAKING)

    # Ensure properties are not rendered at load time
    assert m.physical_properties == {
        "partition_expiration_days": exp.convert(7),
        "creatable_type": exp.maybe_parse(
            "@IF(@model_kind_name = 'FULL', 'TRANSIENT', NULL)", dialect="duckdb"
        ),
    }

    # Substitution occurs at runtime for properties so here these will be unrendered
    assert m.render_physical_properties(
        snapshots={
            m.fqn: snapshot,
        }
    ) == {
        "partition_expiration_days": exp.convert(7),
        "creatable_type": exp.convert("TRANSIENT"),
    }

    assert m.session_properties == {
        "spark.executor.cores": exp.maybe_parse("@IF(@gateway = 'dev', 1, 2)", dialect="duckdb"),
        "spark.executor.memory": "1G",
    }

    assert m.virtual_properties["creatable_type"] == d.MacroVar(this="create_type")

    # Validate rendering of properties
    assert m.render_session_properties(
        snapshots={
            m.fqn: snapshot,
        },
    ) == {
        "spark.executor.cores": exp.convert(2),
        "spark.executor.memory": "1G",
    }

    assert m.render_virtual_properties(
        snapshots={
            m.fqn: snapshot,
        }
    ) == {"creatable_type": exp.convert("SECURE")}

    assert m.render_physical_properties(snapshots={m.fqn: snapshot}) == {
        "partition_expiration_days": exp.convert(7),
        "creatable_type": exp.convert("TRANSIENT"),
    }


@pytest.mark.parametrize(
    "optimize_query, enabled, allow_partials, interval_unit, expected_error",
    [
        ("string", "string", "string", "string", r"^Expected boolean for*"),
        (True, "string", "string", "string", r"^Expected boolean for*"),
        (True, True, "string", "string", r"^Expected boolean for*"),
        (True, True, True, "string", r"^Invalid interval unitr*"),
    ],
)
def test_model_defaults_validations(
    optimize_query, enabled, allow_partials, interval_unit, expected_error
):
    model_defaults = ModelDefaultsConfig(
        optimize_query=optimize_query,
        enabled=enabled,
        allow_partials=allow_partials,
        interval_unit=interval_unit,
    )

    with pytest.raises(ConfigError, match=expected_error):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name test_schema.test_model,
            );
            SELECT a FROM tbl;
            """,
            ),
            defaults=model_defaults.dict(),
        )


def test_model_session_properties(sushi_context):
    assert sushi_context.models['"memory"."sushi"."items"'].session_properties == {
        "string_prop": "some_value",
        "int_prop": 1,
        "float_prop": 1.0,
        "bool_prop": True,
    }
    model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            session_properties (
                'spark.executor.cores' = 2,
                'spark.executor.memory' = '1G',
                some_bool = True,
                some_float = 0.1,
                quoted_identifier = "quoted identifier",
                unquoted_identifier = unquoted_identifier,
            )
        );
        SELECT a FROM tbl;
        """,
            default_dialect="snowflake",
        )
    )

    assert model.session_properties == {
        "spark.executor.cores": 2,
        "spark.executor.memory": "1G",
        "some_bool": True,
        "some_float": 0.1,
        "quoted_identifier": exp.column("quoted identifier", quoted=True),
        "unquoted_identifier": exp.column("unquoted_identifier", quoted=False),
    }

    model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            session_properties (
                'warehouse' = 'test_warehouse'
            )
        );
        SELECT a FROM tbl;
        """,
            default_dialect="snowflake",
        )
    )
    assert model.session_properties == {
        "warehouse": "test_warehouse",
    }

    model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            session_properties (
                'query_label' = [
                    ('key1', 'value1'),
                    ('key2', 'value2')
                ]
            )
        );
        SELECT a FROM tbl;
        """,
            default_dialect="bigquery",
        )
    )
    assert model.session_properties == {
        "query_label": parse_one("[('key1', 'value1'), ('key2', 'value2')]")
    }

    model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            session_properties (
                'query_label' = (
                    ('key1', 'value1')
                )
            )
        );
        SELECT a FROM tbl;
        """,
            default_dialect="bigquery",
        )
    )
    assert model.session_properties == {"query_label": parse_one("(('key1', 'value1'))")}

    with pytest.raises(
        ConfigError,
        match=r"Invalid value for `session_properties.query_label`. Must be an array or tuple.",
    ):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name test_schema.test_model,
                session_properties (
                    'query_label' = 'invalid value'
                )
            );
            SELECT a FROM tbl;
            """,
                default_dialect="bigquery",
            )
        )

    with pytest.raises(
        ConfigError,
        match=r"Invalid entry in `session_properties.query_label`. Must be tuples of string literals with length 2.",
    ):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name test_schema.test_model,
                session_properties (
                    'query_label' = (
                        ('key1', 'value1', 'another_value')
                    )
                )
            );
            SELECT a FROM tbl;
            """,
                default_dialect="bigquery",
            )
        )

    with pytest.raises(
        ConfigError,
        match=r"Invalid entry in `session_properties.query_label`. Must be tuples of string literals with length 2.",
    ):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name test_schema.test_model,
                session_properties (
                    'query_label' = (
                        'some value',
                        'another value',
                        'yet another value',
                    )
                )
            );
            SELECT a FROM tbl;
            """,
                default_dialect="bigquery",
            )
        )


def test_session_properties_authorization_validation():
    model = load_sql_based_model(
        d.parse(
            """
        MODEL (
            name test_schema.test_model,
            session_properties (
                authorization = 'test_user'
            )
        );
        SELECT a FROM tbl;
        """,
            default_dialect="trino",
        )
    )
    assert model.session_properties == {"authorization": "test_user"}

    with pytest.raises(
        ConfigError,
        match=r"Invalid value for `session_properties.authorization`. Must be a string literal.",
    ):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name test_schema.test_model,
                session_properties (
                    authorization = 123
                )
            );
            SELECT a FROM tbl;
            """,
                default_dialect="trino",
            )
        )

    with pytest.raises(
        ConfigError,
        match=r"Invalid value for `session_properties.authorization`. Must be a string literal.",
    ):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name test_schema.test_model,
                session_properties (
                    authorization = some_column
                )
            );
            SELECT a FROM tbl;
            """,
                default_dialect="trino",
            )
        )

    with pytest.raises(
        ConfigError,
        match=r"Invalid value for `session_properties.authorization`. Must be a string literal.",
    ):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name test_schema.test_model,
                session_properties (
                    authorization = CONCAT('user', '_suffix')
                )
            );
            SELECT a FROM tbl;
            """,
                default_dialect="trino",
            )
        )


def test_model_jinja_macro_rendering():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name,
        );

        JINJA_STATEMENT_BEGIN;
        {{ test_package.macro_a() }}
        {{ macro_b() }}
        JINJA_END;

        SELECT 1 AS x;
    """
    )

    jinja_macros = JinjaMacroRegistry(
        packages={
            "test_package": {"macro_a": MacroInfo(definition="macro_a_body", depends_on=[])},
        },
        root_macros={"macro_b": MacroInfo(definition="macro_b_body", depends_on=[])},
        global_objs={"test_int": 1, "test_str": "value"},
    )

    model = load_sql_based_model(expressions, jinja_macros=jinja_macros)
    definition = model.render_definition()

    assert definition[1].sql() == "JINJA_STATEMENT_BEGIN;\nmacro_b_body\nJINJA_END;"
    assert definition[2].sql() == "JINJA_STATEMENT_BEGIN;\nmacro_a_body\nJINJA_END;"


def test_view_model_data_hash():
    view_model_expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind VIEW,
        );
        SELECT 1;
        """
    )
    view_model_hash = load_sql_based_model(view_model_expressions).data_hash

    materialized_view_model_expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind VIEW (
              materialized true
            ),
        );
        SELECT 1;
        """
    )
    materialized_view_model_hash = load_sql_based_model(
        materialized_view_model_expressions
    ).data_hash

    assert view_model_hash != materialized_view_model_hash


def test_view_materialized_partition_by_clustered_by():
    materialized_view_model_expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind VIEW (
              materialized true
            ),
            partitioned_by ds,
            clustered_by a
        );
        SELECT 1;
        """
    )
    materialized_view_model = load_sql_based_model(materialized_view_model_expressions)
    assert materialized_view_model.partitioned_by == [exp.column("ds", quoted=True)]
    assert materialized_view_model.clustered_by == [exp.to_column('"a"')]


def test_view_non_materialized_partition_by():
    view_model_expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind VIEW,
            partitioned_by ds,
        );
        SELECT 1;
        """
    )
    with pytest.raises(ValidationError, match=r".*partitioned_by field cannot be set for VIEW.*"):
        load_sql_based_model(view_model_expressions)


def test_view_non_materialized_clustered_by():
    view_model_expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind VIEW,
            clustered_by ds,
        );
        SELECT 1;
        """
    )
    with pytest.raises(ValidationError, match=r".*clustered_by field cannot be set for VIEW.*"):
        load_sql_based_model(view_model_expressions)


def test_seed_model_data_hash():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
            )
        );
    """
    )
    seed_model = load_sql_based_model(
        expressions, path=Path("./examples/sushi/models/test_model.sql")
    )

    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              csv_settings (
                quotechar = '''',
              )
            )
        );
    """
    )
    new_seed_model = load_sql_based_model(
        expressions, path=Path("./examples/sushi/models/test_model.sql")
    )

    assert seed_model.data_hash != new_seed_model.data_hash


def test_interval_unit_validation():
    assert (
        create_sql_model(
            "a",
            d.parse_one("SELECT a, ds FROM table_a"),
            interval_unit=IntervalUnit.HOUR,
        ).interval_unit
        == IntervalUnit.HOUR
    )

    assert (
        create_sql_model(
            "a",
            d.parse_one("SELECT a, ds FROM table_a"),
            interval_unit="HOUR",
        ).interval_unit
        == IntervalUnit.HOUR
    )

    assert (
        create_sql_model(
            "a",
            d.parse_one("SELECT a, ds FROM table_a"),
            interval_unit=None,
        ).interval_unit_
        is None
    )


def test_scd_type_2_by_time_defaults():
    model_def = d.parse(
        """
        MODEL (
            name db.table,
            kind SCD_TYPE_2 (
                unique_key (COALESCE("ID", '') || '|' || COALESCE("ds", ''), COALESCE("ds", '')),
            ),
        );
        SELECT
            1 as "ID",
            '2020-01-01' as ds,
            '2020-01-01' as test_updated_at,
            '2020-01-01' as test_valid_from,
            '2020-01-01' as test_valid_to
        ;
        """
    )
    scd_type_2_model = load_sql_based_model(model_def)
    assert scd_type_2_model.unique_key == [
        parse_one("""COALESCE("ID", '') || '|' || COALESCE("ds", '')"""),
        parse_one("""COALESCE("ds", '')"""),
    ]
    assert scd_type_2_model.columns_to_types == {
        "ID": exp.DataType.build("int"),
        "ds": exp.DataType.build("varchar"),
        "test_updated_at": exp.DataType.build("varchar"),
        "test_valid_from": exp.DataType.build("varchar"),
        "test_valid_to": exp.DataType.build("varchar"),
        "valid_from": exp.DataType.build("TIMESTAMP"),
        "valid_to": exp.DataType.build("TIMESTAMP"),
    }
    assert scd_type_2_model.managed_columns == {
        "valid_from": exp.DataType.build("TIMESTAMP"),
        "valid_to": exp.DataType.build("TIMESTAMP"),
    }
    assert scd_type_2_model.kind.updated_at_name == exp.column("updated_at", quoted=True)
    assert scd_type_2_model.kind.valid_from_name == exp.column("valid_from", quoted=True)
    assert scd_type_2_model.kind.valid_to_name == exp.column("valid_to", quoted=True)
    assert not scd_type_2_model.kind.updated_at_as_valid_from
    assert scd_type_2_model.kind.is_scd_type_2_by_time
    assert scd_type_2_model.kind.is_scd_type_2
    assert scd_type_2_model.kind.is_materialized
    assert scd_type_2_model.kind.forward_only
    assert scd_type_2_model.kind.disable_restatement


def test_scd_type_2_by_time_overrides():
    model_def = d.parse(
        """
        MODEL (
            name db.table,
            kind SCD_TYPE_2_BY_TIME (
                unique_key ["iD", ds],
                updated_at_name test_updated_at,
                valid_from_name test_valid_from,
                valid_to_name test_valid_to,
                time_data_type TIMESTAMPTZ,
                updated_at_as_valid_from True,
                forward_only False,
                disable_restatement False,
                invalidate_hard_deletes False,
            ),
            dialect snowflake
        );
        SELECT
            1 as "iD",
            '2020-01-01' as ds,
            '2020-01-01' as test_updated_at,
            '2020-01-01' as test_valid_from,
            '2020-01-01' as test_valid_to
        ;
        """
    )
    scd_type_2_model = load_sql_based_model(model_def)
    assert scd_type_2_model.unique_key == [
        exp.column("iD", quoted=True),
        exp.column("DS", quoted=True),
    ]
    assert scd_type_2_model.managed_columns == {
        "TEST_VALID_FROM": exp.DataType.build("TIMESTAMPTZ"),
        "TEST_VALID_TO": exp.DataType.build("TIMESTAMPTZ"),
    }
    assert scd_type_2_model.kind.updated_at_name == exp.column("TEST_UPDATED_AT", quoted=True)
    assert scd_type_2_model.kind.valid_from_name == exp.column("TEST_VALID_FROM", quoted=True)
    assert scd_type_2_model.kind.valid_to_name == exp.column("TEST_VALID_TO", quoted=True)
    assert scd_type_2_model.kind.updated_at_as_valid_from
    assert scd_type_2_model.kind.is_scd_type_2_by_time
    assert scd_type_2_model.kind.is_scd_type_2
    assert scd_type_2_model.kind.is_materialized
    assert not scd_type_2_model.kind.invalidate_hard_deletes
    assert not scd_type_2_model.kind.forward_only
    assert not scd_type_2_model.kind.disable_restatement

    model_kind_dict = scd_type_2_model.kind.dict()
    assert scd_type_2_model.kind == _model_kind_validator(None, model_kind_dict, None)


def test_scd_type_2_by_column_defaults():
    model_def = d.parse(
        """
        MODEL (
            name db.table,
            kind SCD_TYPE_2_BY_COLUMN (
                unique_key "ID",
                columns ["value_to_track"]
            ),
        );
        SELECT
            1 as "ID",
            2 as "value_to_track",
            '2020-01-01' as ds,
        ;
        """
    )
    scd_type_2_model = load_sql_based_model(model_def)
    assert scd_type_2_model.unique_key == [exp.to_column("ID", quoted=True)]
    assert scd_type_2_model.kind.columns == [exp.to_column("value_to_track", quoted=True)]
    assert scd_type_2_model.columns_to_types == {
        "ID": exp.DataType.build("int"),
        "value_to_track": exp.DataType.build("int"),
        "ds": exp.DataType.build("varchar"),
        "valid_from": exp.DataType.build("TIMESTAMP"),
        "valid_to": exp.DataType.build("TIMESTAMP"),
    }
    assert scd_type_2_model.managed_columns == {
        "valid_from": exp.DataType.build("TIMESTAMP"),
        "valid_to": exp.DataType.build("TIMESTAMP"),
    }
    assert scd_type_2_model.kind.valid_from_name == exp.column("valid_from", quoted=True)
    assert scd_type_2_model.kind.valid_to_name == exp.column("valid_to", quoted=True)
    assert not scd_type_2_model.kind.execution_time_as_valid_from
    assert scd_type_2_model.kind.is_scd_type_2_by_column
    assert scd_type_2_model.kind.is_scd_type_2
    assert scd_type_2_model.kind.is_materialized
    assert scd_type_2_model.kind.forward_only
    assert scd_type_2_model.kind.disable_restatement


def test_scd_type_2_by_column_overrides():
    model_def = d.parse(
        """
        MODEL (
            name db.table,
            kind SCD_TYPE_2_BY_COLUMN (
                unique_key ["iD", ds],
                columns "value_to_track",
                valid_from_name test_valid_from,
                valid_to_name test_valid_to,
                execution_time_as_valid_from True,
                time_data_type TIMESTAMPTZ,
                forward_only False,
                disable_restatement False,
                invalidate_hard_deletes False,
                batch_size 1
            ),
        );
        SELECT
            1 as "ID",
            2 as "value_to_track",
            '2020-01-01' as ds,
        ;
        """
    )
    scd_type_2_model = load_sql_based_model(model_def)
    assert scd_type_2_model.unique_key == [
        exp.column("iD", quoted=True),
        exp.column("ds", quoted=True),
    ]
    assert scd_type_2_model.managed_columns == {
        "test_valid_from": exp.DataType.build("TIMESTAMPTZ"),
        "test_valid_to": exp.DataType.build("TIMESTAMPTZ"),
    }
    assert scd_type_2_model.kind.valid_from_name == exp.column("test_valid_from", quoted=True)
    assert scd_type_2_model.kind.valid_to_name == exp.column("test_valid_to", quoted=True)
    assert scd_type_2_model.kind.execution_time_as_valid_from
    assert scd_type_2_model.kind.is_scd_type_2_by_column
    assert scd_type_2_model.kind.is_scd_type_2
    assert scd_type_2_model.kind.is_materialized
    assert scd_type_2_model.kind.time_data_type == exp.DataType.build("TIMESTAMPTZ")
    assert scd_type_2_model.kind.batch_size == 1
    assert not scd_type_2_model.kind.invalidate_hard_deletes
    assert not scd_type_2_model.kind.forward_only
    assert not scd_type_2_model.kind.disable_restatement

    model_kind_dict = scd_type_2_model.kind.dict()
    assert scd_type_2_model.kind == _model_kind_validator(None, model_kind_dict, None)


def test_scd_type_2_python_model() -> None:
    @model(
        "test_scd_type_2_python_model",
        kind=dict(
            name=ModelKindName.SCD_TYPE_2_BY_TIME,
            unique_key="a",
            updated_at_name="b",
            updated_at_as_valid_from=True,
        ),
        columns={"a": "string", "b": "string"},
    )
    def scd_type_2_model(context, **kwargs):
        return pd.DataFrame(
            [
                {
                    "a": "val1",
                    "b": "2024-01-01",
                }
            ]
        )

    python_model = model.get_registry()["test_scd_type_2_python_model"].model(
        module_path=Path("."),
        path=Path("."),
    )

    assert python_model.columns_to_types == {
        "a": exp.DataType.build("string"),
        "b": exp.DataType.build("string"),
        "valid_from": exp.DataType.build("TIMESTAMP"),
        "valid_to": exp.DataType.build("TIMESTAMP"),
    }


@pytest.mark.parametrize(
    "input_columns,expected_columns",
    [
        (
            "col1",
            [exp.to_column("col1", quoted=True)],
        ),
        (
            "[col1]",
            [exp.to_column("col1", quoted=True)],
        ),
        (
            "[col1, col2]",
            [exp.to_column("col1", quoted=True), exp.to_column("col2", quoted=True)],
        ),
        (
            '"col1"',
            [exp.to_column("col1", quoted=True)],
        ),
        (
            '["col1"]',
            [exp.to_column("col1", quoted=True)],
        ),
        ("*", exp.Star()),
    ],
)
def test_check_column_variants(input_columns, expected_columns):
    model_def = d.parse(
        f"""
        MODEL (
            name db.table,
            kind SCD_TYPE_2_BY_COLUMN (
                unique_key "ID",
                columns {input_columns}
            ),
        );
        SELECT 1
        ;
        """
    )
    scd_type_2_model = load_sql_based_model(model_def)
    assert scd_type_2_model.kind.columns == expected_columns


def test_model_dialect_name():
    expressions = d.parse(
        """
        MODEL (
            name `project-1`.`db`.`tbl1`,
            dialect bigquery
        );
        SELECT 1;
        """
    )

    model = load_sql_based_model(expressions)
    assert model.fqn == '"project-1"."db"."tbl1"'

    model = create_external_model(
        "`project-1`.`db`.`tbl1`", columns={"x": "STRING"}, dialect="bigquery"
    )
    assert "name `project-1`.`db`.`tbl1`" in model.render_definition()[0].sql(dialect="bigquery")

    # This used to fail due to the dialect regex picking up `DIALECT_TEST` as the model's dialect
    expressions = d.parse("MODEL(name DIALECT_TEST.foo); SELECT 1")


def test_model_allow_partials():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            allow_partials true,
        );
        SELECT 1;
        """
    )

    model = load_sql_based_model(expressions)

    assert model.allow_partials

    assert "allow_partials TRUE" in model.render_definition()[0].sql()


def test_signals():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            signals [
                (arg = 1),
            ],
        );
        SELECT 1;
        """
    )

    model = load_sql_based_model(expressions)
    assert model.signals[0][1] == {"arg": exp.Literal.number(1)}

    @signal()
    def my_signal(batch):
        return True

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            signals [
                my_signal(arg = 1),
                (
                    table_name := 'table_a',
                    ds := @end_ds,
                ),
                (
                    table_name = 'table_b',
                    ds = @end_ds,
                    hour = @end_hour,
                ),
                (
                    bool_key = True,
                    int_key = 1,
                    float_key = 1.0,
                    string_key = 'string',
                )
            ],
        );
        SELECT 1;
        """
    )

    model = load_sql_based_model(
        expressions,
        signal_definitions={"my_signal": signal.get_registry()["my_signal"]},
    )
    assert model.signals == [
        (
            "my_signal",
            {
                "arg": exp.Literal.number(1),
            },
        ),
        (
            "",
            {
                "table_name": exp.Literal.string("table_a"),
                "ds": d.MacroVar(this="end_ds"),
            },
        ),
        (
            "",
            {
                "table_name": exp.Literal.string("table_b"),
                "ds": d.MacroVar(this="end_ds"),
                "hour": d.MacroVar(this="end_hour"),
            },
        ),
        (
            "",
            {
                "bool_key": exp.true(),
                "int_key": exp.Literal.number(1),
                "float_key": exp.Literal.number(1.0),
                "string_key": exp.Literal.string("string"),
            },
        ),
    ]

    rendered_signals = model.render_signals(start="2023-01-01", end="2023-01-02 15:00:00")
    assert rendered_signals == [
        {"table_name": "table_a", "ds": "2023-01-02"},
        {"table_name": "table_b", "ds": "2023-01-02", "hour": 14},
        {"bool_key": True, "int_key": 1, "float_key": 1.0, "string_key": "string"},
    ]

    assert (
        "signals (MY_SIGNAL(arg := 1), (table_name = 'table_a', ds = @end_ds), (table_name = 'table_b', ds = @end_ds, hour = @end_hour), (bool_key = TRUE, int_key = 1, float_key = 1.0, string_key = 'string'))"
        in model.render_definition()[0].sql()
    )


def test_load_python_model_with_signals():
    @signal()
    def always_true(batch):
        return True

    @model(
        name="model_with_signal",
        kind="full",
        columns={'"COL"': "int"},
        signals=[("always_true", {})],
    )
    def model_with_signal(context, **kwargs):
        return pd.DataFrame([{"COL": 1}])

    models = model.get_registry()["model_with_signal"].models(
        get_variables=lambda _: {},
        path=Path("."),
        module_path=Path("."),
        signal_definitions=signal.get_registry(),
    )
    assert len(models) == 1
    assert models[0].signals == [("always_true", {})]


def test_null_column_type():
    expressions = d.parse(
        """
        MODEL (
          name test_db.test_model,
          columns (
            id INT,
            ds NULL,
          )
        );

        SELECT
          id::INT AS id,
          ds
        FROM x
    """
    )
    model = load_sql_based_model(expressions, dialect="hive")
    assert model.columns_to_types == {
        "ds": exp.DataType.build("null"),
        "id": exp.DataType.build("int"),
    }
    assert not model.annotated


def test_when_matched():
    expressions = d.parse(
        """
        MODEL (
          name db.employees,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key name,
            when_matched WHEN MATCHED THEN UPDATE SET target.salary = COALESCE(source.salary, target.salary)
          )
        );
        SELECT 'name' AS name, 1 AS salary;
    """
    )

    expected_when_matched = "(WHEN MATCHED THEN UPDATE SET `__MERGE_TARGET__`.`salary` = COALESCE(`__MERGE_SOURCE__`.`salary`, `__MERGE_TARGET__`.`salary`))"

    model = load_sql_based_model(expressions, dialect="hive")
    assert model.kind.when_matched.sql(dialect="hive") == expected_when_matched

    model = SqlModel.parse_raw(model.json())
    assert model.kind.when_matched.sql(dialect="hive") == expected_when_matched

    expressions = d.parse(
        """
        MODEL (
          name @{macro_val}.test,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key purchase_order_id,
            when_matched (
              WHEN MATCHED AND source._operation = 1 THEN DELETE
              WHEN MATCHED AND source._operation <> 1 THEN UPDATE SET target.purchase_order_id = 1
            )
          )
        );

        SELECT
          purchase_order_id
        FROM @{macro_val}.upstream
        """
    )

    model = SqlModel.parse_raw(load_sql_based_model(expressions).json())
    assert d.format_model_expressions(model.render_definition()) == (
        """MODEL (
  name @{macro_val}.test,
  kind INCREMENTAL_BY_UNIQUE_KEY (
    unique_key ("purchase_order_id"),
    when_matched (
      WHEN MATCHED AND "__MERGE_SOURCE__"."_operation" = 1 THEN DELETE
      WHEN MATCHED AND "__MERGE_SOURCE__"."_operation" <> 1 THEN UPDATE SET
        "__MERGE_TARGET__"."purchase_order_id" = 1
    ),
    batch_concurrency 1,
    forward_only FALSE,
    disable_restatement FALSE,
    on_destructive_change 'ERROR',
    on_additive_change 'ALLOW'
  )
);

SELECT
  purchase_order_id
FROM @{macro_val}.upstream"""
    )

    @macro()
    def fingerprint_merge(
        evaluator: MacroEvaluator,
        fingerprint_column: exp.Column,
        update_columns: list[exp.Column],
    ) -> exp.Whens:
        fingerprint_evaluation = f"source.{fingerprint_column} <> target.{fingerprint_column}"
        column_update = [f"target.{column} = source.{column}" for column in update_columns]
        return exp.maybe_parse(
            f"WHEN MATCHED AND {fingerprint_evaluation} THEN UPDATE SET {column_update}",
            into=exp.Whens,
        )

    expressions = d.parse(
        """
        MODEL (
          name test,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key purchase_order_id,
            when_matched (@fingerprint_merge(salary, [update_datetime, salary]))
          )
        );

        SELECT
          1 AS purchase_order_id,
          1 AS salary,
          CAST('2020-01-01 12:05:01' AS DATETIME) AS update_datetime
        """
    )

    model = SqlModel.parse_raw(load_sql_based_model(expressions).json())
    assert d.format_model_expressions(model.render_definition()) == (
        """MODEL (
  name test,
  kind INCREMENTAL_BY_UNIQUE_KEY (
    unique_key ("purchase_order_id"),
    when_matched (
      WHEN MATCHED AND "__MERGE_SOURCE__"."salary" <> "__MERGE_TARGET__"."salary" THEN UPDATE SET
        ARRAY('target.update_datetime = source.update_datetime', 'target.salary = source.salary')
    ),
    batch_concurrency 1,
    forward_only FALSE,
    disable_restatement FALSE,
    on_destructive_change 'ERROR',
    on_additive_change 'ALLOW'
  )
);

SELECT
  1 AS purchase_order_id,
  1 AS salary,
  '2020-01-01 12:05:01'::DATETIME AS update_datetime"""
    )


def test_when_matched_multiple():
    expressions = d.parse(
        """
        MODEL (
          name @{schema}.employees,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key name,
            when_matched WHEN MATCHED AND source.x = 1 THEN UPDATE SET target.salary = COALESCE(source.salary, target.salary),
            WHEN MATCHED THEN UPDATE SET target.salary = COALESCE(source.salary, target.salary)

          )
        );
        SELECT 'name' AS name, 1 AS salary;
    """
    )

    expected_when_matched = [
        "WHEN MATCHED AND `__MERGE_SOURCE__`.`x` = 1 THEN UPDATE SET `__MERGE_TARGET__`.`salary` = COALESCE(`__MERGE_SOURCE__`.`salary`, `__MERGE_TARGET__`.`salary`)",
        "WHEN MATCHED THEN UPDATE SET `__MERGE_TARGET__`.`salary` = COALESCE(`__MERGE_SOURCE__`.`salary`, `__MERGE_TARGET__`.`salary`)",
    ]

    model = load_sql_based_model(expressions, dialect="hive", variables={"schema": "db"})
    whens = model.kind.when_matched
    assert len(whens.expressions) == 2
    assert whens.expressions[0].sql(dialect="hive") == expected_when_matched[0]
    assert whens.expressions[1].sql(dialect="hive") == expected_when_matched[1]

    model = SqlModel.parse_raw(model.json())
    whens = model.kind.when_matched
    assert len(whens.expressions) == 2
    assert whens.expressions[0].sql(dialect="hive") == expected_when_matched[0]
    assert whens.expressions[1].sql(dialect="hive") == expected_when_matched[1]


def test_when_matched_merge_filter_multi_part_columns():
    expressions = d.parse(
        """
        MODEL (
          name @{schema}.records_model,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key name,
            when_matched (WHEN MATCHED AND source.record.nested_record.field = 1  THEN UPDATE SET target.repeated_record.sub_repeated_record.sub_field = COALESCE(source.repeated_record.sub_repeated_record.sub_field, target.repeated_record.sub_repeated_record.sub_field),
            WHEN MATCHED THEN UPDATE SET target.repeated_record.sub_repeated_record.sub_field = COALESCE(source.repeated_record.sub_repeated_record.sub_field, target.repeated_record.sub_repeated_record.sub_field)),
            merge_filter source.record.nested_record.field < target.record.nested_record.field AND
            target.repeated_record.sub_repeated_record.sub_field > source.repeated_record.sub_repeated_record.sub_field
          )
        );
        SELECT
            id,
            [STRUCT([STRUCT(sub_field AS sub_field)] AS sub_repeated_record)] AS repeated_record,
            STRUCT(
                STRUCT([2, 3] AS array, field AS field) AS nested_record
            ) AS record
        FROM
            @{schema}.seed_model;
    """
    )

    expected_when_matched = [
        "WHEN MATCHED AND `__MERGE_SOURCE__`.`record`.`nested_record`.`field` = 1 THEN UPDATE SET `__MERGE_TARGET__`.`repeated_record`.`sub_repeated_record`.`sub_field` = COALESCE(`__MERGE_SOURCE__`.`repeated_record`.`sub_repeated_record`.`sub_field`, `__MERGE_TARGET__`.`repeated_record`.`sub_repeated_record`.`sub_field`)",
        "WHEN MATCHED THEN UPDATE SET `__MERGE_TARGET__`.`repeated_record`.`sub_repeated_record`.`sub_field` = COALESCE(`__MERGE_SOURCE__`.`repeated_record`.`sub_repeated_record`.`sub_field`, `__MERGE_TARGET__`.`repeated_record`.`sub_repeated_record`.`sub_field`)",
    ]

    expected_merge_filter = (
        "`__MERGE_SOURCE__`.`record`.`nested_record`.`field` < `__MERGE_TARGET__`.`record`.`nested_record`.`field` AND "
        "`__MERGE_TARGET__`.`repeated_record`.`sub_repeated_record`.`sub_field` > `__MERGE_SOURCE__`.`repeated_record`.`sub_repeated_record`.`sub_field`"
    )

    model = load_sql_based_model(expressions, dialect="bigquery", variables={"schema": "db"})
    whens = model.kind.when_matched
    assert len(whens.expressions) == 2
    assert whens.expressions[0].sql(dialect="bigquery") == expected_when_matched[0]
    assert whens.expressions[1].sql(dialect="bigquery") == expected_when_matched[1]
    assert model.merge_filter.sql(dialect="bigquery") == expected_merge_filter

    model = SqlModel.parse_raw(model.json())
    whens = model.kind.when_matched
    assert len(whens.expressions) == 2
    assert whens.expressions[0].sql(dialect="bigquery") == expected_when_matched[0]
    assert whens.expressions[1].sql(dialect="bigquery") == expected_when_matched[1]
    assert model.merge_filter.sql(dialect="bigquery") == expected_merge_filter


def test_when_matched_normalization() -> None:
    # unquoted should be normalized and quoted
    expressions = d.parse(
        """
        MODEL (
          name test.employees,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key name,
            when_matched (
                WHEN MATCHED THEN UPDATE SET
                    target.key_a = source.key_a,
                    target.key_b = source.key_b,
            )
          )
        );
        SELECT 'name' AS name, 1 AS key_a, 2 AS key_b;
    """
    )
    model = load_sql_based_model(expressions, dialect="snowflake")

    assert isinstance(model.kind, IncrementalByUniqueKeyKind)
    assert isinstance(model.kind.when_matched, exp.Whens)
    first_expression = model.kind.when_matched.expressions[0]
    assert isinstance(first_expression, exp.Expression)
    assert (
        first_expression.sql(dialect="snowflake")
        == 'WHEN MATCHED THEN UPDATE SET "__MERGE_TARGET__"."KEY_A" = "__MERGE_SOURCE__"."KEY_A", "__MERGE_TARGET__"."KEY_B" = "__MERGE_SOURCE__"."KEY_B"'
    )

    # quoted should be preserved
    expressions = d.parse(
        """
        MODEL (
          name test.employees,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key name,
            when_matched (
                WHEN MATCHED THEN UPDATE SET
                    target."kEy_A" = source."kEy_A",
                    target."kEY_b" = source.key_b,
            )
          )
        );
        SELECT 'name' AS name, 1 AS "kEy_A", 2 AS "kEY_b";
    """
    )
    model = load_sql_based_model(expressions, dialect="snowflake")

    assert isinstance(model.kind, IncrementalByUniqueKeyKind)
    assert isinstance(model.kind.when_matched, exp.Whens)
    first_expression = model.kind.when_matched.expressions[0]
    assert isinstance(first_expression, exp.Expression)
    assert (
        first_expression.sql(dialect="snowflake")
        == 'WHEN MATCHED THEN UPDATE SET "__MERGE_TARGET__"."kEy_A" = "__MERGE_SOURCE__"."kEy_A", "__MERGE_TARGET__"."kEY_b" = "__MERGE_SOURCE__"."KEY_B"'
    )


def test_default_catalog_sql(assert_exp_eq):
    """
    This test validates the hashing behavior of the system as it relates to the default catalog.
    The system is not designed to actually support having an engine that doesn't support default catalog
    to start supporting it or the reverse of that. If that did happen then bugs would occur.
    """
    HASH_WITH_CATALOG = "2768215345"

    # Test setting default catalog doesn't change hash if it matches existing logic
    expressions = d.parse(
        """
        MODEL (
            name catalog.db.table
        );
        SELECT x
        FROM catalog.db.source
        """
    )

    model = load_sql_based_model(expressions, default_catalog="catalog")
    assert model.default_catalog == "catalog"
    assert model.name == "catalog.db.table"
    assert model.fqn == '"catalog"."db"."table"'
    assert model.depends_on == {'"catalog"."db"."source"'}

    assert_exp_eq(
        model.render_query(),
        """
        SELECT
          "x" AS "x"
          FROM "catalog"."db"."source" AS "source"
        """,
    )

    assert model.data_hash == HASH_WITH_CATALOG

    expressions = d.parse(
        """
        MODEL (
            name catalog.db.table,
        );
        SELECT x
        FROM catalog.db.source
        """
    )

    model = load_sql_based_model(expressions)
    assert model.default_catalog is None
    assert model.name == "catalog.db.table"
    assert model.fqn == '"catalog"."db"."table"'
    assert model.depends_on == {'"catalog"."db"."source"'}

    assert_exp_eq(
        model.render_query(),
        """
        SELECT
          "x" AS "x"
          FROM "catalog"."db"."source" AS "source"
        """,
    )

    assert model.data_hash == HASH_WITH_CATALOG

    # Test setting default catalog to a different catalog but everything if fully qualified then no hash change
    expressions = d.parse(
        """
        MODEL (
            name catalog.db.table
        );
        SELECT x
        FROM catalog.db.source
        """
    )

    model = load_sql_based_model(expressions, default_catalog="other_catalog")
    assert model.default_catalog == "other_catalog"
    assert model.name == "catalog.db.table"
    assert model.fqn == '"catalog"."db"."table"'
    assert model.depends_on == {'"catalog"."db"."source"'}

    assert_exp_eq(
        model.render_query(),
        """
        SELECT
          "x" AS "x"
          FROM "catalog"."db"."source" AS "source"
        """,
    )

    assert model.data_hash == HASH_WITH_CATALOG

    # test that hash changes if model contains a non-fully-qualified reference
    expressions = d.parse(
        """
        MODEL (
            name catalog.db.table
        );
        SELECT x
        FROM db.source
        """
    )

    model = load_sql_based_model(expressions, default_catalog="other_catalog")
    assert model.default_catalog == "other_catalog"
    assert model.name == "catalog.db.table"
    assert model.fqn == '"catalog"."db"."table"'
    assert model.depends_on == {'"other_catalog"."db"."source"'}

    # The query changed so the hash should change
    assert model.data_hash != HASH_WITH_CATALOG

    # test that hash is the same but the fqn is different so the snapshot is different so this is
    # a new snapshot but with the same hash as before
    expressions = d.parse(
        """
        MODEL (
            name db.table,
        );
        SELECT x
        FROM catalog.db.source
        """
    )

    model = load_sql_based_model(expressions)
    assert model.default_catalog is None
    assert model.name == "db.table"
    assert model.fqn == '"db"."table"'
    assert model.depends_on == {'"catalog"."db"."source"'}

    assert model.data_hash == HASH_WITH_CATALOG

    # This will also have the same hash but the fqn is different so the snapshot is different so this is
    # a new snapshot but with the same hash as before
    expressions = d.parse(
        """
        MODEL (
            name db.table
        );
        SELECT x
        FROM catalog.db.source
        """
    )

    model = load_sql_based_model(expressions, default_catalog="catalog")
    assert model.default_catalog == "catalog"
    assert model.name == "db.table"
    assert model.fqn == '"catalog"."db"."table"'
    assert model.depends_on == {'"catalog"."db"."source"'}

    assert model.data_hash == HASH_WITH_CATALOG

    # Query is different since default catalog does not apply and therefore the hash is different
    expressions = d.parse(
        """
        MODEL (
            name table
        );
        SELECT x
        FROM source
        """
    )

    model = load_sql_based_model(expressions, default_catalog="catalog")
    assert model.default_catalog == "catalog"
    assert model.name == "table"
    assert model.fqn == '"table"'
    assert model.depends_on == {'"source"'}

    assert model.data_hash != HASH_WITH_CATALOG


def test_default_catalog_python():
    HASH_WITH_CATALOG = "2728996410"

    @model(name="db.table", kind="full", columns={'"COL"': "int"})
    def my_model(context, **kwargs):
        context.resolve_table("dependency.table")

    m = model.get_registry()["db.table"].model(
        module_path=Path("."),
        path=Path("."),
    )

    assert m.default_catalog is None
    assert m.name == "db.table"
    assert m.fqn == '"db"."table"'
    assert m.depends_on == {'"dependency"."table"'}

    assert m.data_hash != HASH_WITH_CATALOG

    m = model.get_registry()["db.table"].model(
        module_path=Path("."),
        path=Path("."),
        default_catalog="catalog",
    )

    assert m.default_catalog == "catalog"
    assert m.name == "db.table"
    assert m.fqn == '"catalog"."db"."table"'
    assert m.depends_on == {'"catalog"."dependency"."table"'}

    # This ideally would be `m.data_hash == HASH_WITH_CATALOG`. The reason it is not is
    # because when we hash the python function we make the hash out of the actual logic
    # of the function which means `context.resolve_table("dependency.table")` is used
    # when really is should be `context.resolve_table("catalog.dependency.table")`.
    assert m.data_hash != HASH_WITH_CATALOG

    @model(name="catalog.db.table", kind="full", columns={'"COL"': "int"})
    def my_model(context, **kwargs):
        context.resolve_table("catalog.dependency.table")

    m = model.get_registry()["catalog.db.table"].model(
        module_path=Path("."),
        path=Path("."),
        default_catalog="other_catalog",
    )

    assert m.default_catalog == "other_catalog"
    assert m.name == "catalog.db.table"
    assert m.fqn == '"catalog"."db"."table"'
    assert m.depends_on == {'"catalog"."dependency"."table"'}

    assert m.data_hash == HASH_WITH_CATALOG

    @model(name="catalog.db.table2", kind="full", columns={'"COL"': "int"})
    def my_model(context, **kwargs):
        context.resolve_table("dependency.table")

    m = model.get_registry()["catalog.db.table2"].model(
        module_path=Path("."),
        path=Path("."),
        default_catalog="other_catalog",
    )

    assert m.default_catalog == "other_catalog"
    assert m.name == "catalog.db.table2"
    assert m.fqn == '"catalog"."db"."table2"'
    assert m.depends_on == {'"other_catalog"."dependency"."table"'}

    assert m.data_hash != HASH_WITH_CATALOG

    @model(name="table", kind="full", columns={'"COL"': "int"})
    def my_model(context, **kwargs):
        context.resolve_table("table2")

    m = model.get_registry()["table"].model(
        module_path=Path("."),
        path=Path("."),
        default_catalog="catalog",
    )

    assert m.default_catalog == "catalog"
    assert m.name == "table"
    assert m.fqn == '"table"'
    assert m.depends_on == {'"table2"'}

    assert m.data_hash != HASH_WITH_CATALOG


def test_default_catalog_external_model():
    """
    Since external models fqns are the only thing affected by default catalog, and when they change new snapshots
    are made, the hash will be the same across different names.
    """
    EXPECTED_HASH = "763256265"

    model = create_external_model("db.table", columns={"a": "int", "limit": "int"})
    assert model.default_catalog is None
    assert model.name == "db.table"
    assert model.fqn == '"db"."table"'

    assert model.data_hash == EXPECTED_HASH

    model = create_external_model(
        "db.table", columns={"a": "int", "limit": "int"}, default_catalog="catalog"
    )
    assert model.default_catalog == "catalog"
    assert model.name == "db.table"
    assert model.fqn == '"catalog"."db"."table"'

    assert model.data_hash == EXPECTED_HASH

    model = create_external_model(
        "catalog.db.table", columns={"a": "int", "limit": "int"}, default_catalog="other_catalog"
    )
    assert model.default_catalog == "other_catalog"
    assert model.name == "catalog.db.table"
    assert model.fqn == '"catalog"."db"."table"'

    assert model.data_hash == EXPECTED_HASH

    # Since there is no schema defined, the default physical schema is used which changes the hash
    model = create_external_model(
        "table", columns={"a": "int", "limit": "int"}, default_catalog="catalog"
    )

    assert model.default_catalog == "catalog"
    assert model.name == "table"
    assert model.fqn == '"table"'

    assert model.data_hash != EXPECTED_HASH


def test_user_cannot_set_default_catalog():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            default_catalog some_catalog
        );

        SELECT 1::int AS a, 2::int AS b, 3 AS c, 4 as d;
    """
    )

    with pytest.raises(ConfigError, match="`default_catalog` cannot be set on a per-model basis"):
        load_sql_based_model(expressions)

    with pytest.raises(ConfigError, match="`default_catalog` cannot be set on a per-model basis"):

        @model(name="db.table", kind="full", columns={'"COL"': "int"}, default_catalog="catalog")
        def my_model(context, **kwargs):
            context.resolve_table("dependency.table")


def test_depends_on_default_catalog_python():
    @model(name="some.table", kind="full", columns={'"COL"': "int"}, depends_on={"other.table"})
    def my_model(context, **kwargs):
        context.resolve_table("dependency.table")

    m = model.get_registry()["some.table"].model(
        module_path=Path("."),
        path=Path("."),
        default_catalog="catalog",
    )

    assert m.default_catalog == "catalog"
    assert m.depends_on == {'"catalog"."other"."table"'}


def test_end_date():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE (
                time_column ts,
            ),
            start '2023-01-01',
            end '2023-06-01'
        );

        SELECT 1::int AS a, 2::int AS b, now::timestamp as ts
        """
    )
    model = load_sql_based_model(expressions)

    assert model.start == "2023-01-01"
    assert model.end == "2023-06-01"
    assert model.interval_unit == IntervalUnit.DAY

    with pytest.raises(ValidationError, match=".*Start date.+can't be greater than end date.*"):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name db.table,
                kind INCREMENTAL_BY_TIME_RANGE (
                    time_column ts,
                ),
                start '2024-01-01',
                end '2023-06-01'
            );

            SELECT 1::int AS a, 2::int AS b, now::timestamp as ts
            """
            )
        )


def test_end_no_start():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE (
                time_column ts,
            ),
            end '2023-06-01'
        );

        SELECT 1::int AS a, 2::int AS b, now::timestamp as ts
        """
    )
    with pytest.raises(ConfigError, match="Must define a start date if an end date is defined"):
        load_sql_based_model(expressions)
    load_sql_based_model(expressions, defaults={"start": "2023-01-01"})


def test_variables():
    @macro()
    def test_macro_var(evaluator) -> exp.Expression:
        return exp.convert(evaluator.var("TEST_VAR_D") + 10)

    expressions = parse(
        """
        MODEL(
            name test_model,
            kind FULL,
        );

        SELECT
          @VAR('TEST_VAR_A') AS a,
          @VAR('test_var_b', 'default_value') AS b,
          @VAR('test_var_c') AS c,
          @TEST_MACRO_VAR() AS d,
          @'foo_@{test_var_e}' AS e,
          @SQL(foo_@{test_var_f}) AS f,
          'foo_@{test_var_unused}' AS g
    """,
        default_dialect="bigquery",
    )

    model = load_sql_based_model(
        expressions,
        variables={
            "test_var_a": "test_value",
            "test_var_d": 1,
            "test_var_e": 4,
            "test_var_f": 5,
            "test_var_unused": 2,
        },
    )
    assert model.python_env[c.SQLMESH_VARS] == Executable.value(
        {"test_var_a": "test_value", "test_var_d": 1, "test_var_e": 4, "test_var_f": 5}
    )
    assert (
        model.render_query().sql(dialect="bigquery")
        == "SELECT 'test_value' AS `a`, 'default_value' AS `b`, NULL AS `c`, 11 AS `d`, 'foo_4' AS `e`, `foo_5` AS `f`, 'foo_@{test_var_unused}' AS `g`"
    )

    with pytest.raises(ConfigError, match=r"Macro VAR requires at least one argument.*"):
        expressions = parse(
            """
            MODEL(
                name test_model,
            );

            SELECT @VAR() AS a;
        """,
            default_dialect="bigquery",
        )
        load_sql_based_model(expressions)

    with pytest.raises(
        ConfigError, match=r"The variable name must be a string literal, '123' was given instead.*"
    ):
        expressions = parse(
            """
            MODEL(
                name test_model,
            );

            SELECT @VAR(123) AS a;
        """,
            default_dialect="bigquery",
        )
        load_sql_based_model(expressions)

    with pytest.raises(
        ConfigError,
        match=r"The variable name must be a string literal, '@VAR_NAME' was given instead.*",
    ):
        expressions = parse(
            """
            MODEL(
                name test_model,
            );

            @DEF(VAR_NAME, 'var_name');
            SELECT @VAR(@VAR_NAME) AS a;
        """,
            default_dialect="bigquery",
        )
        load_sql_based_model(expressions)


def test_named_variable_macros() -> None:
    model = load_sql_based_model(
        parse(
            """
        MODEL(name sushi.test_gateway_macro);
        @DEF(overridden_var, 'overridden_value');
        SELECT @gateway AS gateway, @TEST_VAR_A AS test_var_a, @overridden_var AS overridden_var
        """
        ),
        variables={
            c.GATEWAY: "in_memory",
            "test_var_a": "test_value",
            "test_var_unused": "unused",
            "overridden_var": "initial_value",
        },
    )

    assert model.python_env[c.SQLMESH_VARS] == Executable.value(
        {c.GATEWAY: "in_memory", "test_var_a": "test_value", "overridden_var": "initial_value"},
        sort_root_dict=True,
    )
    assert (
        model.render_query_or_raise().sql()
        == "SELECT 'in_memory' AS \"gateway\", 'test_value' AS \"test_var_a\", 'overridden_value' AS \"overridden_var\""
    )


def test_variables_in_templates() -> None:
    model = load_sql_based_model(
        parse(
            """
        MODEL(name sushi.test_gateway_macro);
        @DEF(overridden_var, overridden_value);
        SELECT 'gateway' AS col_@gateway, 'test_var_a' AS @{test_var_a}_col, 'overridden_var' AS col_@{overridden_var}_col
        """
        ),
        variables={
            c.GATEWAY: "in_memory",
            "test_var_a": "test_value",
            "test_var_unused": "unused",
            "overridden_var": "initial_value",
        },
    )

    assert model.python_env[c.SQLMESH_VARS] == Executable.value(
        {c.GATEWAY: "in_memory", "test_var_a": "test_value", "overridden_var": "initial_value"},
        sort_root_dict=True,
    )
    assert (
        model.render_query_or_raise().sql()
        == "SELECT 'gateway' AS \"col_in_memory\", 'test_var_a' AS \"test_value_col\", 'overridden_var' AS \"col_overridden_value_col\""
    )

    model = load_sql_based_model(
        parse(
            """
        MODEL(name sushi.test_gateway_macro);
        @DEF(overridden_var, overridden_value);
        SELECT 'combo' AS col_@{test_var_a}_@{overridden_var}_col_@gateway
        """
        ),
        variables={
            c.GATEWAY: "in_memory",
            "test_var_a": "test_value",
            "test_var_unused": "unused",
            "overridden_var": "initial_value",
        },
    )

    assert model.python_env[c.SQLMESH_VARS] == Executable.value(
        {c.GATEWAY: "in_memory", "test_var_a": "test_value", "overridden_var": "initial_value"},
        sort_root_dict=True,
    )
    assert (
        model.render_query_or_raise().sql()
        == "SELECT 'combo' AS \"col_test_value_overridden_value_col_in_memory\""
    )

    model = load_sql_based_model(
        parse(
            """
        MODEL(
          name @{some_var}.bar,
          dialect snowflake
        );

        SELECT 1 AS c
        """
        ),
        variables={
            "some_var": "foo",
        },
    )

    assert model.name == "foo.bar"


def test_variables_jinja():
    expressions = parse(
        """
        MODEL(
            name test_model,
            kind FULL,
        );

        JINJA_QUERY_BEGIN;
        SELECT '{{ var('TEST_VAR_A') }}' AS a, '{{ var('test_var_b', 'default_value') }}' AS b, '{{ var('test_var_c') }}' AS c, {{ test_macro_var() }} AS d;
        JINJA_END;
    """,
        default_dialect="bigquery",
    )

    jinja_macros = JinjaMacroRegistry(
        root_macros={
            "test_macro_var": MacroInfo(
                definition="{% macro test_macro_var() %}{{ var('test_var_d') + 10 }}{% endmacro %}",
                depends_on=[],
            )
        },
    )

    model = load_sql_based_model(
        expressions,
        variables={"test_var_a": "test_value", "test_var_d": 1, "test_var_unused": 2},
        jinja_macros=jinja_macros,
    )
    assert model.python_env[c.SQLMESH_VARS] == Executable.value(
        {"test_var_a": "test_value", "test_var_d": 1}
    )
    assert (
        model.render_query().sql(dialect="bigquery")
        == "SELECT 'test_value' AS `a`, 'default_value' AS `b`, 'None' AS `c`, 11 AS `d`"
    )


def test_variables_python_model(mocker: MockerFixture) -> None:
    @model(
        "foo_@{bar}",
        kind="full",
        columns={"a": "string", "b": "string", "c": "string"},
    )
    def model_with_variables(context, **kwargs):
        return pd.DataFrame(
            [
                {
                    "a": context.var("TEST_VAR_A"),
                    "b": context.var("test_var_b", "default_value"),
                    "c": context.var("test_var_c"),
                }
            ]
        )

    python_model = model.get_registry()["foo_@{bar}"].model(
        module_path=Path("."),
        path=Path("."),
        variables={"test_var_a": "test_value", "test_var_unused": 2, "bar": "suffix"},
    )

    assert python_model.name == "foo_suffix"
    assert python_model.python_env[c.SQLMESH_VARS] == Executable.value({"test_var_a": "test_value"})

    context = ExecutionContext(mocker.Mock(), {}, None, None)
    df = list(python_model.render(context=context))[0]
    assert df.to_dict(orient="records") == [{"a": "test_value", "b": "default_value", "c": None}]


def test_load_external_model_python(sushi_context) -> None:
    @model(
        "test_load_external_model_python",
        columns={"customer_id": "int", "zip": "str"},
        kind={"name": ModelKindName.FULL},
    )
    def external_model_python(context, **kwargs):
        demographics_table = context.resolve_table("memory.raw.demographics")
        return context.fetchdf(
            exp.select("customer_id", "zip").from_(demographics_table),
        )

    python_model = model.get_registry()["test_load_external_model_python"].model(
        module_path=Path("."),
        path=Path("."),
    )

    context = ExecutionContext(sushi_context.engine_adapter, sushi_context.snapshots, None, None)
    df = list(python_model.render(context=context))[0]

    assert df.to_dict(orient="records") == [{"customer_id": 1, "zip": "00000"}]


def test_variables_python_sql_model(mocker: MockerFixture) -> None:
    @model(
        "test_variables_python_model_@{bar}",
        is_sql=True,
        kind="full",
        columns={"a": "string", "b": "string", "c": "string"},
    )
    def model_with_variables(evaluator, **kwargs):
        return exp.select(
            exp.convert(evaluator.var("TEST_VAR_A")).as_("a"),
            exp.convert(evaluator.var("test_var_b", "default_value")).as_("b"),
            exp.convert(evaluator.var("test_var_c")).as_("c"),
        )

    python_sql_model = model.get_registry()["test_variables_python_model_@{bar}"].model(
        module_path=Path("."),
        path=Path("."),
        variables={"test_var_a": "test_value", "test_var_unused": 2, "bar": "suffix"},
    )

    assert python_sql_model.name == "test_variables_python_model_suffix"
    assert python_sql_model.python_env[c.SQLMESH_VARS] == Executable.value(
        {"test_var_a": "test_value"}
    )

    context = ExecutionContext(mocker.Mock(), {}, None, None)
    query = list(python_sql_model.render(context=context))[0]
    assert (
        query.sql()
        == """SELECT 'test_value' AS "a", 'default_value' AS "b", NULL AS "c" """.strip()
    )


def test_macros_python_model(mocker: MockerFixture) -> None:
    @model(
        "foo_macro_model_@{bar}",
        columns={"a": "string"},
        kind=dict(name=ModelKindName.INCREMENTAL_BY_TIME_RANGE, time_column="@{time_col}"),
        stamp="@{stamp}",
        cron="@some_cron_var",
        owner="@IF(@gateway = 'dev', @{dev_owner}, @{prod_owner})",
        enabled="@IF(@gateway = 'dev', True, False)",
        start="@IF(@gateway = 'dev', '1 month ago', '2024-01-01')",
        partitioned_by=[
            d.parse_one("DATETIME_TRUNC(@{time_col}, MONTH)"),
        ],
    )
    def model_with_macros(context, **kwargs):
        return pd.DataFrame(
            [
                {
                    "a": context.var("TEST_VAR_A"),
                }
            ]
        )

    python_model = model.get_registry()["foo_macro_model_@{bar}"].model(
        module_path=Path("."),
        path=Path("."),
        variables={
            "test_var_a": "test_value",
            "gateway": "prod",
            "bar": "suffix",
            "dev_owner": "dv_1",
            "prod_owner": "pr_1",
            "stamp": "bump",
            "time_col": "a",
            "some_cron_var": "@daily",
        },
    )

    assert python_model.name == "foo_macro_model_suffix"
    assert python_model.python_env[c.SQLMESH_VARS] == Executable.value({"test_var_a": "test_value"})
    assert not python_model.enabled
    assert python_model.start == "2024-01-01"
    assert python_model.owner == "pr_1"
    assert python_model.stamp == "bump"
    assert python_model.time_column.column == exp.column("a", quoted=True)
    assert python_model.partitioned_by[0].sql() == 'DATETIME_TRUNC("a", MONTH)'
    assert python_model.cron == "@daily"

    context = ExecutionContext(mocker.Mock(), {}, None, None)
    df = list(python_model.render(context=context))[0]
    assert df.to_dict(orient="records") == [{"a": "test_value"}]


def test_macros_python_sql_model(mocker: MockerFixture) -> None:
    @macro()
    def end_date_macro(evaluator: MacroEvaluator, var: bool):
        return f"@IF({var} = False, '1 day ago', '2025-01-01 12:00:00')"

    @model(
        "test_macros_python_model_@{bar}",
        is_sql=True,
        kind="full",
        cron="@daily",
        columns={"a": "string"},
        enabled="@IF(@gateway = 'dev', True, False)",
        start="@IF(@gateway = 'dev', '1 month ago', '2024-01-01')",
        end="@end_date_macro(@{global_var})",
        owner="@IF(@gateway = 'dev', @{dev_owner}, @{prod_owner})",
        stamp="@{stamp}",
        tags=["@{tag1}", "@{tag2}"],
        description="'Model desc @{test_}'",
    )
    def model_with_macros(evaluator, **kwargs):
        return exp.select(
            exp.convert(evaluator.var("TEST_VAR_A")).as_("a"),
        )

    python_sql_model = model.get_registry()["test_macros_python_model_@{bar}"].model(
        module_path=Path("."),
        path=Path("."),
        macros=macro.get_registry(),
        variables={
            "test_var_a": "test_value",
            "test_var_unused": 2,
            "bar": "suffix",
            "gateway": "dev",
            "global_var": False,
            "dev_owner": "dv_1",
            "prod_owner": "pr_1",
            "stamp": "bump",
            "time_col": "a",
            "tag1": "tag__1",
            "tag2": "tag__2",
        },
    )

    assert python_sql_model.name == "test_macros_python_model_suffix"
    assert python_sql_model.python_env[c.SQLMESH_VARS] == Executable.value(
        {"test_var_a": "test_value"}
    )

    assert python_sql_model.enabled
    assert python_sql_model.start == "1 month ago"
    assert python_sql_model.end == "1 day ago"
    assert python_sql_model.owner == "dv_1"
    assert python_sql_model.stamp == "bump"
    assert python_sql_model.description == "Model desc @{test_}"
    assert python_sql_model.tags == ["tag__1", "tag__2"]

    context = ExecutionContext(mocker.Mock(), {}, None, None)
    query = list(python_sql_model.render(context=context))[0]
    assert query.sql() == """SELECT 'test_value' AS "a" """.strip()


def test_unrendered_macros_sql_model(mocker: MockerFixture) -> None:
    model = load_sql_based_model(
        parse(
            """
            MODEL (
              name db.employees,
              kind INCREMENTAL_BY_UNIQUE_KEY (
                unique_key @{key},
                merge_filter source.id > 0 and target.updated_at < @end_ds and source.updated_at > @start_ds and @merge_filter_var
              ),
              cron '@daily',
              allow_partials @IF(@gateway = 'dev', True, False),
              physical_properties (
                location1 = @'s3://bucket/prefix/@{schema_name}/@{table_name}',
                location2 = @IF(@gateway = 'dev', @'hdfs://@{catalog_name}/@{schema_name}/dev/@{table_name}', @'s3://prod/@{table_name}'),
                foo = @physical_var
              ),
              virtual_properties (
                creatable_type = @{create_type},
                bar = @virtual_var,
              ),
              session_properties (
                'spark.executor.cores' = @IF(@gateway = 'dev', 1, 2),
                'spark.executor.memory' = '1G',
                baz = @session_var
              ),
            );

            SELECT * FROM src;
        """
        ),
        variables={
            "gateway": "dev",
            "key": "a",  # Not included in python_env because kind is rendered at load time
            "create_type": "'SECURE'",
            "merge_filter_var": True,
            "physical_var": "bla",
            "virtual_var": "blb",
            "session_var": "blc",
        },
    )

    assert model.python_env[c.SQLMESH_VARS] == Executable.value(
        {
            "gateway": "dev",
            "create_type": "'SECURE'",
            "merge_filter_var": True,
            "physical_var": "bla",
            "virtual_var": "blb",
            "session_var": "blc",
        },
        sort_root_dict=True,
    )

    assert "location1" in model.physical_properties
    assert "location2" in model.physical_properties

    # The properties will stay unrendered at load time
    assert model.session_properties == {
        "spark.executor.cores": exp.maybe_parse("@IF(@gateway = 'dev', 1, 2)"),
        "spark.executor.memory": "1G",
        "baz": exp.maybe_parse("@session_var"),
    }
    assert model.virtual_properties["creatable_type"] == exp.maybe_parse("@{create_type}")

    assert (
        model.physical_properties["location1"].sql()
        == "@'s3://bucket/prefix/@{schema_name}/@{table_name}'"
    )
    assert (
        model.physical_properties["location2"].sql()
        == "@IF(@gateway = 'dev', @'hdfs://@{catalog_name}/@{schema_name}/dev/@{table_name}', @'s3://prod/@{table_name}')"
    )

    # merge_filter will stay unrendered as well
    assert model.unique_key[0] == exp.column("a", quoted=True)
    assert (
        t.cast(exp.Expression, model.merge_filter).sql()
        == '"__MERGE_SOURCE__"."id" > 0 AND "__MERGE_TARGET__"."updated_at" < @end_ds AND "__MERGE_SOURCE__"."updated_at" > @start_ds AND @merge_filter_var'
    )


def test_unrendered_macros_python_model(mocker: MockerFixture) -> None:
    @model(
        "test_unrendered_macros_python_model_@{bar}",
        is_sql=True,
        kind=dict(
            name=ModelKindName.INCREMENTAL_BY_UNIQUE_KEY,
            unique_key="@{key}",
            merge_filter="source.id > 0 and target.updated_at < @end_ds and source.updated_at > @start_ds and @merge_filter_var",
        ),
        cron="@daily",
        columns={"a": "string"},
        allow_partials="@IF(@gateway = 'dev', True, False)",
        physical_properties=dict(
            location1="@'s3://bucket/prefix/@{schema_name}/@{table_name}'",
            location2="@IF(@gateway = 'dev', @'hdfs://@{catalog_name}/@{schema_name}/dev/@{table_name}', @'s3://prod/@{table_name}')",
            foo="@physical_var",
        ),
        virtual_properties={"creatable_type": "@{create_type}", "bar": "@virtual_var"},
        session_properties={
            "spark.executor.cores": "@IF(@gateway = 'dev', 1, 2)",
            "spark.executor.memory": "1G",
            "baz": "@session_var",
        },
    )
    def model_with_macros(evaluator, **kwargs):
        return exp.select(
            exp.convert(evaluator.var("TEST_VAR_A")).as_("a"),
        )

    python_sql_model = model.get_registry()["test_unrendered_macros_python_model_@{bar}"].model(
        module_path=Path("."),
        path=Path("."),
        macros=macro.get_registry(),
        variables={
            "test_var_a": "test_value",
            "bar": "suffix",
            "gateway": "dev",
            "key": "a",
            "create_type": "'SECURE'",
            "merge_filter_var": True,
            "physical_var": "bla",
            "virtual_var": "blb",
            "session_var": "blc",
        },
    )

    assert python_sql_model.name == "test_unrendered_macros_python_model_suffix"
    assert python_sql_model.python_env[c.SQLMESH_VARS] == Executable.value(
        {
            "test_var_a": "test_value",
            "gateway": "dev",
            "create_type": "'SECURE'",
            "merge_filter_var": True,
            "physical_var": "bla",
            "virtual_var": "blb",
            "session_var": "blc",
        },
        sort_root_dict=True,
    )
    assert python_sql_model.enabled

    context = ExecutionContext(mocker.Mock(), {}, None, None)
    query = list(python_sql_model.render(context=context))[0]
    assert query.sql() == """SELECT 'test_value' AS "a" """.strip()
    assert python_sql_model.allow_partials

    assert "location1" in python_sql_model.physical_properties
    assert "location2" in python_sql_model.physical_properties

    # The properties will stay unrendered at load time
    assert python_sql_model.session_properties == {
        "spark.executor.cores": exp.maybe_parse("@IF(@gateway = 'dev', 1, 2)"),
        "spark.executor.memory": "1G",
        "baz": exp.maybe_parse("@session_var"),
    }
    assert python_sql_model.virtual_properties["creatable_type"] == exp.maybe_parse(
        "@{create_type}"
    )

    assert (
        python_sql_model.physical_properties["location1"].sql()
        == "@'s3://bucket/prefix/@{schema_name}/@{table_name}'"
    )
    assert (
        python_sql_model.physical_properties["location2"].sql()
        == "@IF(@gateway = 'dev', @'hdfs://@{catalog_name}/@{schema_name}/dev/@{table_name}', @'s3://prod/@{table_name}')"
    )

    # merge_filter will stay unrendered as well
    assert python_sql_model.unique_key[0] == exp.column("a", quoted=True)
    assert (
        python_sql_model.merge_filter.sql()
        == '"__MERGE_SOURCE__"."id" > 0 AND "__MERGE_TARGET__"."updated_at" < @end_ds AND "__MERGE_SOURCE__"."updated_at" > @start_ds AND @merge_filter_var'
    )


def test_columns_python_sql_model() -> None:
    @model(
        "test_columns_python_model",
        is_sql=True,
        kind="full",
        columns={"d": "Date", "s": "String", "dt": "DateTime"},
    )
    def model_with_columns(evaluator, **kwargs):
        return exp.select("*").from_("fake")

    python_sql_model = model.get_registry()["test_columns_python_model"].model(
        module_path=Path("."),
        path=Path("."),
    )

    columns_to_types = python_sql_model.columns_to_types

    assert columns_to_types is not None
    assert isinstance(columns_to_types["d"], exp.DataType)
    assert columns_to_types["d"].this == exp.DataType.Type.DATE
    assert isinstance(columns_to_types["s"], exp.DataType)
    assert columns_to_types["s"].this == exp.DataType.Type.TEXT
    assert isinstance(columns_to_types["dt"], exp.DataType)
    assert columns_to_types["dt"].this == exp.DataType.Type.DATETIME


def test_named_variables_python_model(mocker: MockerFixture) -> None:
    mocker.patch("sqlmesh.core.model.decorator.model._registry", {})

    @model(
        "test_named_variables_python_model",
        kind="full",
        columns={"a": "string", "b": "string", "c": "string"},
    )
    def model_with_named_variables(
        context, start: TimeLike, test_var_a: str, test_var_b: t.Optional[str] = None, **kwargs
    ):
        return pd.DataFrame(
            [{"a": test_var_a, "b": test_var_b, "start": start.strftime("%Y-%m-%d")}]  # type: ignore
        )

    python_model = model.get_registry()["test_named_variables_python_model"].model(
        module_path=Path("."),
        path=Path("."),
        # Passing `start` in variables to make sure that built-in arguments can't be overridden.
        variables={
            "test_var_a": "test_value",
            "test_var_unused": 2,
            "start": "2024-01-01",
        },
    )

    assert python_model.python_env[c.SQLMESH_VARS] == Executable.value(
        {"test_var_a": "test_value", "start": "2024-01-01"}, sort_root_dict=True
    )

    context = ExecutionContext(mocker.Mock(), {}, None, None)
    df = list(python_model.render(context=context))[0]
    assert df.to_dict(orient="records") == [{"a": "test_value", "b": None, "start": to_ds(c.EPOCH)}]


def test_named_variables_kw_only_python_model(mocker: MockerFixture) -> None:
    mocker.patch("sqlmesh.core.model.decorator.model._registry", {})

    @model(
        "test_named_variables_python_model",
        kind="full",
        columns={"a": "string"},
    )
    def model_with_named_kw_only_variables(
        context, start: TimeLike, *, test_var_a: str = "", **kwargs: t.Any
    ):
        return pd.DataFrame([{"a": test_var_a}])

    python_model = model.get_registry()["test_named_variables_python_model"].model(
        module_path=Path("."),
        path=Path("."),
        variables={"test_var_a": "test_value"},
    )

    assert python_model.python_env[c.SQLMESH_VARS] == Executable.value({"test_var_a": "test_value"})

    context = ExecutionContext(mocker.Mock(), {}, None, None)
    df = list(python_model.render(context=context))[0]
    assert df.to_dict(orient="records") == [{"a": "test_value"}]


def test_gateway_macro() -> None:
    model = load_sql_based_model(
        parse(
            """
        MODEL(name sushi.test_gateway_macro);
        SELECT @gateway AS gateway
        """
        ),
        variables={c.GATEWAY: "in_memory"},
    )

    assert model.python_env[c.SQLMESH_VARS] == Executable.value({c.GATEWAY: "in_memory"})
    assert model.render_query_or_raise().sql() == "SELECT 'in_memory' AS \"gateway\""

    @macro()
    def macro_uses_gateway(evaluator) -> exp.Expression:
        return exp.convert(evaluator.gateway + "_from_macro")

    model = load_sql_based_model(
        parse(
            """
        MODEL(name sushi.test_gateway_macro);
        SELECT @macro_uses_gateway() AS gateway_from_macro
        """
        ),
        variables={c.GATEWAY: "in_memory"},
    )

    assert model.python_env[c.SQLMESH_VARS] == Executable.value({c.GATEWAY: "in_memory"})
    assert (
        model.render_query_or_raise().sql()
        == "SELECT 'in_memory_from_macro' AS \"gateway_from_macro\""
    )


def test_gateway_macro_jinja() -> None:
    model = load_sql_based_model(
        parse(
            """
        MODEL(name sushi.test_gateway_macro_jinja);
        JINJA_QUERY_BEGIN;
        SELECT '{{ gateway() }}' AS gateway_jinja;
        JINJA_END;
        """
        ),
        variables={c.GATEWAY: "in_memory"},
    )

    assert model.python_env[c.SQLMESH_VARS] == Executable.value({c.GATEWAY: "in_memory"})
    assert model.render_query_or_raise().sql() == "SELECT 'in_memory' AS \"gateway_jinja\""


def test_gateway_python_model(mocker: MockerFixture) -> None:
    @model(
        "test_gateway_python_model",
        kind="full",
        columns={"gateway_python": "string"},
    )
    def model_with_variables(context, **kwargs):
        return pd.DataFrame([{"gateway_python": context.gateway + "_from_python"}])

    python_model = model.get_registry()["test_gateway_python_model"].model(
        module_path=Path("."),
        path=Path("."),
        variables={c.GATEWAY: "in_memory"},
    )

    assert python_model.python_env[c.SQLMESH_VARS] == Executable.value({c.GATEWAY: "in_memory"})

    context = ExecutionContext(mocker.Mock(), {}, None, None)
    df = list(python_model.render(context=context))[0]
    assert df.to_dict(orient="records") == [{"gateway_python": "in_memory_from_python"}]


@pytest.mark.parametrize("dialect", ["spark", "trino"])
def test_view_render_no_quote_identifiers(dialect: str) -> None:
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind VIEW,
        );
        SELECT a, b, c FROM source_table;
        """
    )
    model = load_sql_based_model(expressions, dialect=dialect)
    assert (
        model.render_query_or_raise().sql(dialect=dialect)
        == "SELECT a AS a, b AS b, c AS c FROM source_table AS source_table"
    )


@pytest.mark.parametrize(
    "dialect,kind",
    [
        ("spark", "FULL"),
        ("trino", "FULL"),
        ("duckdb", "VIEW"),
        ("duckdb", "FULL"),
    ],
)
def test_render_quote_identifiers(dialect: str, kind: str) -> None:
    expressions = d.parse(
        f"""
        MODEL (
            name db.table,
            kind {kind},
        );
        SELECT a, b, c FROM source_table;
        """
    )
    model = load_sql_based_model(expressions, dialect=dialect)
    assert (
        model.render_query_or_raise().sql(dialect="duckdb")
        == 'SELECT "a" AS "a", "b" AS "b", "c" AS "c" FROM "source_table" AS "source_table"'
    )


def test_this_model() -> None:
    expressions = d.parse(
        """
        MODEL (
            name `project-1.table`,
            dialect bigquery,
        );

        JINJA_STATEMENT_BEGIN;
        VACUUM {{ this_model }} TO 'a';
        JINJA_END;

        JINJA_QUERY_BEGIN;
        SELECT '{{ this_model }}' as x
        JINJA_END;

        JINJA_STATEMENT_BEGIN;
        VACUUM {{ this_model }} TO 'b';
        JINJA_END;
        """
    )
    model = load_sql_based_model(expressions)

    assert (
        model.render_query_or_raise().sql(dialect="bigquery")
        == """SELECT '`project-1`.`table`' AS `x`"""
    )

    assert (
        model.render_pre_statements()[0].sql(dialect="bigquery")
        == """VACUUM `project-1`.`table` TO 'a'"""
    )
    assert (
        model.render_post_statements()[0].sql(dialect="bigquery")
        == """VACUUM `project-1`.`table` TO 'b'"""
    )

    snapshot = Snapshot.from_node(model, nodes={})

    assert (
        model.render_query_or_raise(
            start="2020-01-01",
            snapshots={snapshot.name: snapshot},
        ).sql(dialect="bigquery")
        == """SELECT '`project-1`.`table`' AS `x`"""
    )

    snapshot.categorize_as(SnapshotChangeCategory.BREAKING)

    assert (
        model.render_query_or_raise(
            start="2020-01-01",
            snapshots={snapshot.name: snapshot},
        ).sql(dialect="bigquery")
        == f"SELECT '`sqlmesh__project-1`.`project_1__table__{snapshot.version}`' AS `x`"
    )

    @macro()
    def this_model_resolves_to_quoted_table(evaluator):
        this_model = evaluator.locals.get("this_model")
        if not this_model:
            return True

        this_snapshot = evaluator.get_snapshot("db.table")
        if this_snapshot and this_snapshot.version:
            # If the table wasn't quoted, we'd break the `sqlmesh_DB` reference by
            # normalizing it twice
            expected_name = f'"sqlmesh__DB"."DB__TABLE__{this_snapshot.version}"'
        else:
            expected_name = '"DB"."TABLE"'

        return not this_model or (
            isinstance(this_model, exp.Table)
            and this_model.sql(dialect=evaluator.dialect, comments=False) == expected_name
            and evaluator.this_model == expected_name
        )

    expressions = d.parse(
        """
        MODEL (name db.table, dialect snowflake);

        SELECT
          1 AS col,
          @this_model_resolves_to_quoted_table() AS this_model_resolves_to_quoted_table;

        CREATE TABLE db.other AS SELECT * FROM @this_model AS x;
        """
    )
    model = load_sql_based_model(expressions)

    expected_post = d.parse('CREATE TABLE "DB"."OTHER" AS SELECT * FROM "DB"."TABLE" AS "X";')
    assert model.render_post_statements() == expected_post

    snapshot = Snapshot.from_node(model, nodes={})
    assert (
        model.render_query_or_raise(snapshots={snapshot.name: snapshot}, start="2020-01-01").sql(
            dialect="snowflake"
        )
        == 'SELECT 1 AS "COL", TRUE AS "THIS_MODEL_RESOLVES_TO_QUOTED_TABLE"'
    )

    snapshot.categorize_as(SnapshotChangeCategory.BREAKING)
    assert (
        model.render_query_or_raise(snapshots={snapshot.name: snapshot}, start="2021-01-01").sql(
            dialect="snowflake"
        )
        == 'SELECT 1 AS "COL", TRUE AS "THIS_MODEL_RESOLVES_TO_QUOTED_TABLE"'
    )


def test_macros_in_physical_properties(make_snapshot):
    expressions = d.parse(
        """
        MODEL (
            name test.test_model,
            kind FULL,
            physical_properties (
                location1 = @resolve_template('s3://bucket/prefix/@{schema_name}/@{table_name}'),
                location2 = @IF(
                    @gateway = 'dev',
                    @resolve_template('hdfs://@{catalog_name}/@{schema_name}/dev/@{table_name}'),
                    @resolve_template('s3://prod/@{table_name}')
                ),
                sort_order = @IF(@gateway = 'prod', 'desc', 'asc'),
                conditional_prop = @IF(@gateway == 'prod', 'PROD_PROP', NULL)
            )
        );

        SELECT 1;
        """
    )

    model = load_sql_based_model(
        expressions, variables={"gateway": "dev"}, default_catalog="unit_test"
    )

    assert model.name == "test.test_model"
    assert "location1" in model.physical_properties
    assert "location2" in model.physical_properties
    assert "sort_order" in model.physical_properties
    assert "conditional_prop" in model.physical_properties

    # load time is a no-op
    assert isinstance(model.physical_properties["location1"], d.MacroFunc)
    assert isinstance(model.physical_properties["location2"], d.MacroFunc)
    assert isinstance(model.physical_properties["sort_order"], d.MacroFunc)
    assert isinstance(model.physical_properties["conditional_prop"], d.MacroFunc)

    # substitution occurs at runtime
    snapshot = make_snapshot(model)
    snapshot.categorize_as(SnapshotChangeCategory.BREAKING)

    rendered_physical_properties = model.render_physical_properties(
        snapshots={model.fqn: snapshot},  # to trigger @this_model generation
        runtime_stage=RuntimeStage.CREATING,
        python_env=model.python_env,
    )

    assert (
        rendered_physical_properties["location1"].text("this")
        == f"s3://bucket/prefix/sqlmesh__test/test__test_model__{snapshot.version}"
    )
    assert (
        rendered_physical_properties["location2"].text("this")
        == f"hdfs://unit_test/sqlmesh__test/dev/test__test_model__{snapshot.version}"
    )
    assert rendered_physical_properties["sort_order"].text("this") == "asc"

    # the conditional_prop will be disabled for "dev" gateway
    assert "conditional_prop" not in rendered_physical_properties


def test_macros_in_model_statement(sushi_context, assert_exp_eq):
    @macro()
    def session_properties(evaluator, value):
        return exp.Property(
            this=exp.var("session_properties"),
            value=exp.convert([exp.convert("foo").eq(exp.var(f"bar_{value}"))]),
        )

    expressions = d.parse(
        """
        MODEL (
            name @{gateway}__@{gateway}.test_model,
            kind INCREMENTAL_BY_TIME_RANGE (
                time_column @{time_column}

            ),
            start @IF(@gateway = 'test_gateway', '2023-01-01', '2024-01-02'),
            @session_properties(baz)
        );

        SELECT a, b UNION SELECT c, c
        """
    )

    model = load_sql_based_model(
        expressions, variables={"gateway": "test_gateway", "time_column": "a"}
    )
    assert model.name == "test_gateway__test_gateway.test_model"
    assert model.time_column
    assert model.time_column.column == exp.column("a", quoted=True)
    assert model.start == "2023-01-01"
    assert model.session_properties == {"foo": exp.column("bar_baz", quoted=False)}


def test_macro_references_in_audits():
    @macro()
    def zero_value(evaluator: MacroEvaluator) -> int:
        return 0

    @macro()
    def min_value(evaluator: MacroEvaluator) -> int:
        return 1

    @macro()
    def not_loaded_macro(evaluator: MacroEvaluator) -> int:
        return 10

    @macro()
    def max_value(evaluator: MacroEvaluator) -> int:
        return 1000

    audit_expression = parse(
        """
        AUDIT (
    name assert_max_value,
    );
    SELECT *
    FROM @this_model
    WHERE
    id > @max_value;
    """
    )

    not_zero_audit = parse(
        """
        AUDIT (
    name assert_not_zero,
    );
    SELECT *
    FROM @this_model
    WHERE
    id = @zero_value;
    """
    )

    model_expression = d.parse(
        """
        MODEL (
            name db.audit_model,
            audits (assert_max_value, assert_positive_ids),
        );
        SELECT 1 as id;

        AUDIT (
    name assert_positive_ids,
    );
    SELECT *
    FROM @this_model
    WHERE
    id < @min_value;
    """
    )

    audits = {
        "assert_max_value": load_audit(audit_expression, dialect="duckdb"),
        "assert_not_zero": load_audit(not_zero_audit, dialect="duckdb"),
    }
    model_defaults = ModelDefaultsConfig(dialect="duckdb", audits=["assert_not_zero"])

    model = load_sql_based_model(
        model_expression,
        defaults=model_defaults.dict(),
        audit_definitions=audits,
    )

    assert len(model.audits) == 3
    audits_with_args = model.audits_with_args
    assert len(audits_with_args) == 3
    assert len(model.python_env) == 3
    assert model.audits == [
        ("assert_not_zero", {}),
        ("assert_max_value", {}),
        ("assert_positive_ids", {}),
    ]
    assert isinstance(audits_with_args[0][0], ModelAudit)
    assert isinstance(audits_with_args[1][0], ModelAudit)
    assert isinstance(audits_with_args[2][0], ModelAudit)
    assert isinstance(model.python_env["min_value"], Executable)
    assert isinstance(model.python_env["max_value"], Executable)
    assert isinstance(model.python_env["zero_value"], Executable)
    assert "not_loaded_macro" not in model.python_env


def test_python_model_dialect():
    model._dialect = "snowflake"

    @model(
        name="a",
        kind=IncrementalByTimeRangeKind(time_column=TimeColumn(column="x", format="YYMMDD")),
        columns={},
    )
    def test(context, **kwargs):
        return None

    m = model.get_registry()["a"].model(
        module_path=Path("."),
        path=Path("."),
        dialect="snowflake",
    )

    assert m.time_column.column.sql() == '"X"'
    assert m.time_column.format == "%y%m%d"

    @model(
        name="b",
        kind=IncrementalByTimeRangeKind(time_column="y"),
        columns={},
    )
    def test(context, **kwargs):
        return None

    m = model.get_registry()["b"].model(
        module_path=Path("."),
        path=Path("."),
        dialect="snowflake",
    )

    assert m.time_column.column.sql() == '"Y"'
    assert m.time_column.format == "%Y-%m-%d"

    # column type parseable by default dialect: no error
    model._dialect = "clickhouse"

    @model("good", columns={'"COL"': "DateTime64(9)"})
    def a_model(context):
        pass

    # column type not parseable by default dialect and no explicit dialect: error
    model._dialect = "snowflake"

    with pytest.raises(ParseError, match="No expression was parsed from 'DateTime64\\(9\\)'"):

        @model("bad", columns={'"COL"': "DateTime64(9)"})
        def a_model(context):
            pass

    # column type not parseable by default dialect and explicit dialect specified: no error
    @model("good", columns={'"COL"': "DateTime64(9)"}, dialect="clickhouse")
    def a_model(context):
        pass

    model._dialect = None


def test_jinja_runtime_stage(assert_exp_eq):
    expressions = d.parse(
        """
        MODEL (
            name test.jinja
        );

        JINJA_QUERY_BEGIN;

        SELECT '{{ runtime_stage }}' as a, {{ runtime_stage == 'loading' }} as b

        JINJA_END;
        """
    )

    model = load_sql_based_model(expressions)
    assert_exp_eq(model.render_query(), '''SELECT 'loading' as "a", TRUE as "b"''')


def test_forward_only_on_destructive_change_config() -> None:
    # global default to ALLOW for non-incremental models
    config = Config(model_defaults=ModelDefaultsConfig(dialect="duckdb"))
    context = Context(config=config)

    expressions = d.parse(
        """
        MODEL (
            name memory.db.table,
            kind FULL,
        );
        SELECT a, b, c FROM source_table;
        """
    )
    model = load_sql_based_model(expressions, defaults=config.model_defaults.dict())
    context.upsert_model(model)
    context_model = context.get_model("memory.db.table")
    assert context_model.on_destructive_change.is_allow

    # global default to ERROR for incremental models
    config = Config(model_defaults=ModelDefaultsConfig(dialect="duckdb"))
    context = Context(config=config)

    expressions = d.parse(
        """
        MODEL (
            name memory.db.table,
            kind INCREMENTAL_BY_TIME_RANGE (
                time_column c,
                forward_only True
            ),
        );
        SELECT a, b, c FROM source_table;
        """
    )
    model = load_sql_based_model(expressions, defaults=config.model_defaults.dict())
    context.upsert_model(model)
    context_model = context.get_model("memory.db.table")
    assert context_model.on_destructive_change.is_error

    # WARN specified in model definition, overrides incremental model sqlmesh default ERROR
    config = Config(model_defaults=ModelDefaultsConfig(dialect="duckdb"))
    context = Context(config=config)

    expressions = d.parse(
        """
        MODEL (
            name memory.db.table,
            kind INCREMENTAL_BY_TIME_RANGE (
                time_column c,
                forward_only True,
                on_destructive_change warn
            ),
        );
        SELECT a, b, c FROM source_table;
        """
    )
    model = load_sql_based_model(expressions, defaults=config.model_defaults.dict())
    context.upsert_model(model)
    context_model = context.get_model("memory.db.table")
    assert context_model.on_destructive_change.is_warn

    # WARN specified as model default, overrides incremental model sqlmesh default ERROR
    config = Config(
        model_defaults=ModelDefaultsConfig(dialect="duckdb", on_destructive_change="warn")
    )
    context = Context(config=config)

    expressions = d.parse(
        """
        MODEL (
            name memory.db.table,
            kind INCREMENTAL_BY_TIME_RANGE (
                time_column c,
                forward_only True
            ),
        );
        SELECT a, b, c FROM source_table;
        """
    )
    model = load_sql_based_model(expressions, defaults=config.model_defaults.dict())
    context.upsert_model(model)
    context_model = context.get_model("memory.db.table")
    assert context_model.on_destructive_change.is_warn

    # WARN specified as model default, does not override non-incremental sqlmesh default ALLOW
    config = Config(
        model_defaults=ModelDefaultsConfig(dialect="duckdb", on_destructive_change="warn")
    )
    context = Context(config=config)

    expressions = d.parse(
        """
        MODEL (
            name memory.db.table,
            kind FULL,
        );
        SELECT a, b, c FROM source_table;
        """
    )
    model = load_sql_based_model(expressions, defaults=config.model_defaults.dict())
    context.upsert_model(model)
    context_model = context.get_model("memory.db.table")
    assert context_model.on_destructive_change.is_allow


def test_model_meta_on_additive_change_property() -> None:
    """Test that ModelMeta has on_additive_change property that works like on_destructive_change."""
    from sqlmesh.core.model.kind import IncrementalByTimeRangeKind, OnAdditiveChange
    from sqlmesh.core.model.meta import ModelMeta

    # Test incremental model with on_additive_change=ERROR
    incremental_kind = IncrementalByTimeRangeKind(
        time_column="c",
        forward_only=True,
        on_additive_change=OnAdditiveChange.ERROR,
    )
    model_meta = ModelMeta(name="test_model", kind=incremental_kind)
    assert model_meta.on_additive_change == OnAdditiveChange.ERROR

    # Test incremental model with on_additive_change=WARN
    incremental_kind = IncrementalByTimeRangeKind(
        time_column="c",
        forward_only=True,
        on_additive_change=OnAdditiveChange.WARN,
    )
    model_meta = ModelMeta(name="test_model", kind=incremental_kind)
    assert model_meta.on_additive_change == OnAdditiveChange.WARN

    # Test incremental model with default on_additive_change (should be ALLOW)
    incremental_kind = IncrementalByTimeRangeKind(
        time_column="c",
        forward_only=True,
    )
    model_meta = ModelMeta(name="test_model", kind=incremental_kind)
    assert model_meta.on_additive_change == OnAdditiveChange.ALLOW

    incremental_kind = IncrementalByTimeRangeKind(
        time_column="c",
        forward_only=True,
        on_additive_change=OnAdditiveChange.IGNORE,
    )
    model_meta = ModelMeta(name="test_model", kind=incremental_kind)
    assert model_meta.on_additive_change == OnAdditiveChange.IGNORE


def test_incremental_by_partition(sushi_context, assert_exp_eq):
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_PARTITION,
            partitioned_by [a],
        );

        SELECT a, b
        """
    )
    model = load_sql_based_model(expressions)
    assert model.kind.is_incremental_by_partition
    assert not model.kind.disable_restatement

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_PARTITION (
                disable_restatement false
            ),
            partitioned_by [a],
        );

        SELECT a, b
        """
    )
    model = load_sql_based_model(expressions)
    assert model.kind.is_incremental_by_partition
    assert not model.kind.disable_restatement

    with pytest.raises(
        ValidationError,
        match=r".*partitioned_by field is required for INCREMENTAL_BY_PARTITION models.*",
    ):
        expressions = d.parse(
            """
            MODEL (
                name db.table,
                kind INCREMENTAL_BY_PARTITION,
            );

            SELECT a, b
            """
        )
        load_sql_based_model(expressions)

    with pytest.raises(
        ConfigError,
        match=r".*Do not specify the `forward_only` configuration key.*",
    ):
        expressions = d.parse(
            """
            MODEL (
                name db.table,
                kind INCREMENTAL_BY_PARTITION (
                    forward_only true
                ),
            );

            SELECT a, b
            """
        )
        load_sql_based_model(expressions)


@pytest.mark.parametrize(
    ["model_def", "path", "expected_name"],
    [
        [
            """dialect duckdb,""",
            """models/test_schema/test_model.sql,""",
            "test_schema.test_model",
        ],
        [
            """dialect duckdb,""",
            """models/test_model.sql,""",
            "test_model",
        ],
        [
            """dialect duckdb,""",
            """models/inventory/db/test_schema/test_model.sql,""",
            "db.test_schema.test_model",
        ],
        ["""name test_model,""", """models/schema/test_model.sql,""", "test_model"],
    ],
)
def test_model_table_name_inference(
    sushi_context: Context, model_def: str, path: str, expected_name: str
):
    model = load_sql_based_model(
        d.parse(
            f"""
        MODEL (
            {model_def}
        );
        SELECT a FROM tbl;
        """,
            default_dialect="duckdb",
        ),
        path=Path(f"$root/{path}"),
        infer_names=True,
    )
    assert model.name == expected_name


@pytest.mark.parametrize(
    ["path", "expected_name"],
    [
        [
            """models/test_schema/test_model.py""",
            "test_schema.test_model",
        ],
        [
            """models/inventory/db/test_schema/test_model.py""",
            "db.test_schema.test_model",
        ],
    ],
)
def test_python_model_name_inference(tmp_path: Path, path: str, expected_name: str) -> None:
    init_example_project(tmp_path, engine_type="duckdb")
    config = Config(
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
        model_naming=NameInferenceConfig(infer_names=True),
    )

    foo_py_file = tmp_path / path
    foo_py_file.parent.mkdir(parents=True, exist_ok=True)
    foo_py_file.write_text("""from sqlmesh import model
@model(
    columns={'"COL"': "int"},
)
def my_model(context, **kwargs):
    pass""")
    context = Context(paths=tmp_path, config=config)
    assert context.get_model(expected_name).name == expected_name
    assert isinstance(context.get_model(expected_name), PythonModel)


def test_python_model_name_inference_multiple_models(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb")
    config = Config(
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
        model_naming=NameInferenceConfig(infer_names=True),
    )

    path_a = tmp_path / "models/test_schema/test_model_a.py"
    path_b = tmp_path / "models/test_schema/test_model_b.py"

    model_payload = """from sqlmesh import model
@model(
    columns={'"COL"': "int"},
)
def my_model(context, **kwargs):
    pass"""

    path_a.parent.mkdir(parents=True, exist_ok=True)
    path_a.write_text(model_payload)
    path_b.write_text(model_payload)

    context = Context(paths=tmp_path, config=config)
    assert context.get_model("test_schema.test_model_a").name == "test_schema.test_model_a"
    assert context.get_model("test_schema.test_model_b").name == "test_schema.test_model_b"


def test_custom_kind():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind CUSTOM (
                materialization 'MyTestStrategy',
                forward_only true,
                disable_restatement true,
                materialization_properties (
                  'key_a' = 'value_a',
                  key_b = 2,
                  'key_c' = true,
                  'key_d' = 1.23,
                ),
                batch_size 1,
                batch_concurrency 2,
                lookback 3,
            )
        );

        SELECT a, b
        """
    )

    with pytest.raises(
        ConfigError, match=r"Materialization strategy with name 'MyTestStrategy' was not found.*"
    ):
        model = load_sql_based_model(expressions)
        model.validate_definition()

    class MyTestStrategy(CustomMaterialization):
        pass

    model = load_sql_based_model(expressions)
    assert model.kind.is_custom

    kind = t.cast(CustomKind, model.kind)
    assert kind.disable_restatement
    assert kind.forward_only
    assert kind.materialization == "MyTestStrategy"
    assert kind.materialization_properties == {
        "key_a": "value_a",
        "key_b": 2,
        "key_c": True,
        "key_d": 1.23,
    }
    assert kind.batch_size == 1
    assert kind.batch_concurrency == 2
    assert kind.lookback == 3

    assert (
        kind.to_expression().sql()
        == """CUSTOM (
materialization 'MyTestStrategy',
materialization_properties ('key_a' = 'value_a', key_b = 2, 'key_c' = TRUE, 'key_d' = 1.23),
forward_only TRUE,
disable_restatement TRUE,
batch_size 1,
batch_concurrency 2,
lookback 3
)"""
    )


def test_custom_kind_lookback_property():
    """Test that CustomKind's lookback property is correctly accessed via ModelMeta.lookback.

    This test verifies the fix for issue #5268 where CustomKind models were not respecting
    the lookback parameter because the isinstance check for _IncrementalBy failed.
    """

    # Test 1: CustomKind with lookback = 3
    class MyTestStrategy(CustomMaterialization):
        pass

    expressions = d.parse(
        """
        MODEL (
            name db.custom_table,
            kind CUSTOM (
                materialization 'MyTestStrategy',
                lookback 3
            )
        );
        SELECT a, b FROM upstream
        """
    )

    model = load_sql_based_model(expressions)
    assert model.kind.is_custom

    # Verify that the kind itself has lookback = 3
    kind = t.cast(CustomKind, model.kind)
    assert kind.lookback == 3

    # The bug: model.lookback should return 3, but with the old implementation
    # using isinstance(self.kind, _IncrementalBy), it would return 0
    assert model.lookback == 3, "CustomKind lookback not accessible via model.lookback property"

    # Test 2: CustomKind without lookback (should default to 0)
    expressions_no_lookback = d.parse(
        """
        MODEL (
            name db.custom_table_no_lookback,
            kind CUSTOM (
                materialization 'MyTestStrategy'
            )
        );
        SELECT a, b FROM upstream
        """
    )

    model_no_lookback = load_sql_based_model(expressions_no_lookback)
    assert model_no_lookback.lookback == 0

    # Test 3: Ensure IncrementalByTimeRangeKind still works correctly
    incremental_expressions = d.parse(
        """
        MODEL (
            name db.incremental_table,
            kind INCREMENTAL_BY_TIME_RANGE (
                time_column ds,
                lookback 5
            )
        );
        SELECT ds, a, b FROM upstream
        """
    )

    incremental_model = load_sql_based_model(incremental_expressions)
    assert incremental_model.lookback == 5


def test_time_column_format_in_custom_kind():
    class TimeColumnCustomKind(CustomKind):  # type: ignore[no-untyped-def]
        _time_column: TimeColumn

        @model_validator(mode="after")
        def _validate(self):
            self._time_column = TimeColumn.create(
                self.materialization_properties.get("time_column"), self.dialect
            )

        @property
        def time_column(self):
            return self._time_column

    class TimeColumnMaterialization(CustomMaterialization[TimeColumnCustomKind]):
        NAME = "time_column_custom_strategy"

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind CUSTOM (
                materialization 'time_column_custom_strategy',
                materialization_properties (
                  time_column = ts
                ),
            ),
            dialect duckdb
        );

        SELECT a, b, '2020-01-01' as ts
        """
    )

    model = load_sql_based_model(expressions, time_column_format="%d-%m-%Y")
    assert isinstance(model.kind, TimeColumnCustomKind)
    assert model.kind.time_column.column == exp.to_column("ts", quoted=True)
    assert model.kind.time_column.format == "%d-%m-%Y"
    assert model.kind.dialect == "duckdb"
    assert "dialect" not in json.loads(
        model.kind.json()
    )  # dialect should not be serialized against the kind

    # explicit time_column format within the model
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind CUSTOM (
                materialization 'time_column_custom_strategy',
                materialization_properties (
                  time_column = (ts, '%Y-%m-%d')
                ),
            )
        );

        SELECT a, b, '2020-01-01' as ts
        """
    )

    model = load_sql_based_model(expressions, time_column_format="%d-%m-%Y")
    assert model.time_column.format == "%Y-%m-%d"


def test_model_kind_to_expression():
    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column a,
            ),
        );
        SELECT a, b
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """INCREMENTAL_BY_TIME_RANGE (
time_column ("a", '%Y-%m-%d'),
partition_by_time_column TRUE,
forward_only FALSE,
disable_restatement FALSE,
on_destructive_change 'ERROR',
on_additive_change 'ALLOW'
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column a,
                batch_size 1,
                batch_concurrency 2,
                lookback 3,
                forward_only TRUE,
                disable_restatement TRUE,
                on_destructive_change WARN,
            ),
        );
        SELECT a, b
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """INCREMENTAL_BY_TIME_RANGE (
time_column ("a", '%Y-%m-%d'),
partition_by_time_column TRUE,
batch_size 1,
batch_concurrency 2,
lookback 3,
forward_only TRUE,
disable_restatement TRUE,
on_destructive_change 'WARN',
on_additive_change 'ALLOW'
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_UNIQUE_KEY(
                unique_key a,
            ),
        );
        SELECT a, b
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """INCREMENTAL_BY_UNIQUE_KEY (
unique_key ("a"),
batch_concurrency 1,
forward_only FALSE,
disable_restatement FALSE,
on_destructive_change 'ERROR',
on_additive_change 'ALLOW'
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_UNIQUE_KEY(
                unique_key a,
                when_matched WHEN MATCHED THEN UPDATE SET target.b = COALESCE(source.b, target.b)
            ),
        );
        SELECT a, b
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """INCREMENTAL_BY_UNIQUE_KEY (
unique_key ("a"),
when_matched (WHEN MATCHED THEN UPDATE SET "__MERGE_TARGET__"."b" = COALESCE("__MERGE_SOURCE__"."b", "__MERGE_TARGET__"."b")),
batch_concurrency 1,
forward_only FALSE,
disable_restatement FALSE,
on_destructive_change 'ERROR',
on_additive_change 'ALLOW'
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name db.table,
                kind INCREMENTAL_BY_UNIQUE_KEY(
                    unique_key a,
                    when_matched WHEN MATCHED AND source.x = 1 THEN UPDATE SET target.b = COALESCE(source.b, target.b),
                    WHEN MATCHED THEN UPDATE SET target.b = COALESCE(source.b, target.b)
                ),
            );
            SELECT a, b
            """
            )
        )
        .kind.to_expression()
        .sql()
        == """INCREMENTAL_BY_UNIQUE_KEY (
unique_key ("a"),
when_matched (WHEN MATCHED AND "__MERGE_SOURCE__"."x" = 1 THEN UPDATE SET "__MERGE_TARGET__"."b" = COALESCE("__MERGE_SOURCE__"."b", "__MERGE_TARGET__"."b") WHEN MATCHED THEN UPDATE SET "__MERGE_TARGET__"."b" = COALESCE("__MERGE_SOURCE__"."b", "__MERGE_TARGET__"."b")),
batch_concurrency 1,
forward_only FALSE,
disable_restatement FALSE,
on_destructive_change 'ERROR',
on_additive_change 'ALLOW'
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_PARTITION,
            partitioned_by ["a"],
        );
        SELECT a, b
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """INCREMENTAL_BY_PARTITION (
forward_only TRUE,
disable_restatement FALSE,
on_destructive_change 'ERROR',
on_additive_change 'ALLOW'
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
            )
        );
        """
            ),
            path=Path("./examples/sushi/models/test_model.sql"),
        )
        .kind.to_expression()
        .sql()
        == """SEED (
path '../seeds/waiter_names.csv',
batch_size 1000
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind SCD_TYPE_2_BY_TIME (
                unique_key [a, b]
            )
        );
        SELECT a, b
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """SCD_TYPE_2_BY_TIME (
updated_at_name "updated_at",
updated_at_as_valid_from FALSE,
unique_key ("a", "b"),
valid_from_name "valid_from",
valid_to_name "valid_to",
invalidate_hard_deletes FALSE,
time_data_type TIMESTAMP,
forward_only TRUE,
disable_restatement TRUE,
on_destructive_change 'ERROR',
on_additive_change 'ALLOW'
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind SCD_TYPE_2_BY_COLUMN (
                unique_key [a, b],
                columns [b]
            )
        );
        SELECT a, b, c
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """SCD_TYPE_2_BY_COLUMN (
columns ("b"),
execution_time_as_valid_from FALSE,
unique_key ("a", "b"),
valid_from_name "valid_from",
valid_to_name "valid_to",
invalidate_hard_deletes FALSE,
time_data_type TIMESTAMP,
forward_only TRUE,
disable_restatement TRUE,
on_destructive_change 'ERROR',
on_additive_change 'ALLOW'
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind SCD_TYPE_2_BY_COLUMN (
                unique_key [a, b],
                columns *
            )
        );
        SELECT a, b, c
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """SCD_TYPE_2_BY_COLUMN (
columns *,
execution_time_as_valid_from FALSE,
unique_key ("a", "b"),
valid_from_name "valid_from",
valid_to_name "valid_to",
invalidate_hard_deletes FALSE,
time_data_type TIMESTAMP,
forward_only TRUE,
disable_restatement TRUE,
on_destructive_change 'ERROR',
on_additive_change 'ALLOW'
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind FULL
        );
        SELECT a, b, c
        """
            )
        )
        .kind.to_expression()
        .sql()
        == "FULL"
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind VIEW
        );
        SELECT a, b, c
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """VIEW (
materialized FALSE
)"""
    )

    assert (
        load_sql_based_model(
            d.parse(
                """
        MODEL (
            name db.table,
            kind VIEW (materialized true)
        );
        SELECT a, b, c
        """
            )
        )
        .kind.to_expression()
        .sql()
        == """VIEW (
materialized TRUE
)"""
    )


def test_bad_model_kind():
    with pytest.raises(
        SQLMeshError,
        match=f"Model kind specified as 'BAD_KIND', but that is not a valid model kind.\n\nPlease specify one of {', '.join(ModelKindName)}.",
    ):
        d.parse(
            """
        MODEL (
            name db.table,
            kind BAD_KIND
        );
        SELECT a, b
        """
        )


def test_merge_filter():
    expressions = d.parse(
        """
        MODEL (
          name db.employees,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key name,
            merge_filter source.salary > 0
          )
        );
        SELECT 'name' AS name, 1 AS salary;
    """
    )

    expected_incremental_predicate = f"`{MERGE_SOURCE_ALIAS}`.`salary` > 0"

    model = load_sql_based_model(expressions, dialect="hive")
    assert model.kind.merge_filter.sql(dialect="hive") == expected_incremental_predicate

    model = SqlModel.parse_raw(model.json())
    assert model.kind.merge_filter.sql(dialect="hive") == expected_incremental_predicate
    assert model.dialect == "hive"

    expressions = d.parse(
        """
        MODEL (
          name db.test,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key purchase_order_id,
            when_matched (
              WHEN MATCHED AND source._operation = 1 THEN DELETE
              WHEN MATCHED AND source._operation <> 1 THEN UPDATE SET target.purchase_order_id = 1
            ),
            merge_filter (
                source.ds > (SELECT MAX(ds) FROM db.test) AND
                source.ds > @start_ds AND
                source._operation <> 1 AND
                target.start_date > date_add(current_date, interval 7 day)
            )
          )
        );

        SELECT
          purchase_order_id,
          start_date
        FROM db.upstream
        """
    )

    model = SqlModel.parse_raw(load_sql_based_model(expressions, dialect="duckdb").json())
    assert d.format_model_expressions(model.render_definition(), dialect=model.dialect) == (
        f"""MODEL (
  name db.test,
  dialect duckdb,
  kind INCREMENTAL_BY_UNIQUE_KEY (
    unique_key ("purchase_order_id"),
    when_matched (
      WHEN MATCHED AND "{MERGE_SOURCE_ALIAS}"."_operation" = 1 THEN DELETE
      WHEN MATCHED AND "{MERGE_SOURCE_ALIAS}"."_operation" <> 1 THEN UPDATE SET
        "{MERGE_TARGET_ALIAS}"."purchase_order_id" = 1
    ),
    merge_filter (
      "{MERGE_SOURCE_ALIAS}"."ds" > (
        SELECT
          MAX("ds")
        FROM "db"."test"
      )
      AND "{MERGE_SOURCE_ALIAS}"."ds" > @start_ds
      AND "{MERGE_SOURCE_ALIAS}"."_operation" <> 1
      AND "{MERGE_TARGET_ALIAS}"."start_date" > CURRENT_DATE + INTERVAL '7' DAY
    ),
    batch_concurrency 1,
    forward_only FALSE,
    disable_restatement FALSE,
    on_destructive_change 'ERROR',
    on_additive_change 'ALLOW'
  )
);

SELECT
  purchase_order_id,
  start_date
FROM db.upstream"""
    )

    rendered_merge_filters = model.render_merge_filter(start="2023-01-01", end="2023-01-02")
    assert (
        rendered_merge_filters.sql(dialect="hive")
        == "(`__MERGE_SOURCE__`.`ds` > (SELECT MAX(`ds`) FROM `db`.`test`) AND `__MERGE_SOURCE__`.`ds` > '2023-01-01' AND `__MERGE_SOURCE__`.`_operation` <> 1 AND `__MERGE_TARGET__`.`start_date` > CURRENT_DATE + INTERVAL '7' DAY)"
    )


def test_merge_filter_normalization():
    # unquoted gets normalized and quoted
    expressions = d.parse(
        """
        MODEL (
          name db.employees,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key name,
            merge_filter source.salary > 0
          )
        );
        SELECT 'name' AS name, 1 AS salary;
    """
    )

    model = load_sql_based_model(expressions, dialect="snowflake")
    assert model.merge_filter.sql(dialect="snowflake") == '"__MERGE_SOURCE__"."SALARY" > 0'

    # quoted gets preserved
    expressions = d.parse(
        """
        MODEL (
          name db.employees,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key name,
            merge_filter source."SaLArY" > 0
          )
        );
        SELECT 'name' AS name, 1 AS "SaLArY";
    """
    )

    model = load_sql_based_model(expressions, dialect="snowflake")
    assert model.merge_filter.sql(dialect="snowflake") == '"__MERGE_SOURCE__"."SaLArY" > 0'


def test_merge_filter_macro():
    @macro()
    def predicate(
        evaluator: MacroEvaluator,
        cluster_column: exp.Column,
    ) -> exp.Expression:
        return parse_one(f"source.{cluster_column} > dateadd(day, -7, target.{cluster_column})")

    expressions = d.parse(
        """
        MODEL (
          name db.incremental_model,
          kind INCREMENTAL_BY_UNIQUE_KEY (
            unique_key id,
            merge_filter @predicate(update_datetime) and target.update_datetime > @start_dt
          ),
          clustered_by update_datetime
        );
        SELECT id, update_datetime FROM db.test_model;
    """
    )

    unrendered_merge_filter = f"""@predicate("UPDATE_DATETIME") AND "{MERGE_TARGET_ALIAS}"."UPDATE_DATETIME" > @start_dt"""
    expected_merge_filter = (
        f"""\"{MERGE_SOURCE_ALIAS}"."UPDATE_DATETIME" > DATEADD(DAY, -7, "{MERGE_TARGET_ALIAS}"."UPDATE_DATETIME") """
        f"""AND "{MERGE_TARGET_ALIAS}"."UPDATE_DATETIME" > CAST('2023-01-01 15:00:00+00:00' AS TIMESTAMPTZ)"""
    )

    model = load_sql_based_model(expressions, dialect="snowflake")
    assert model.kind.merge_filter.sql(dialect=model.dialect) == unrendered_merge_filter

    model = SqlModel.parse_raw(model.json())
    assert model.kind.merge_filter.sql(dialect=model.dialect) == unrendered_merge_filter

    rendered_merge_filters = model.render_merge_filter(start="2023-01-01 15:00:00")
    assert rendered_merge_filters.sql(dialect=model.dialect) == expected_merge_filter


@pytest.mark.parametrize(
    "metadata_only",
    [True, False],
)
def test_macro_func_hash(mocker: MockerFixture, metadata_only: bool):
    mocker.patch("sqlmesh.core.macros.macro._registry", {})

    @macro(metadata_only=metadata_only)
    def noop(evaluator) -> None:
        return None

    expressions = d.parse(
        """
        MODEL (
            name db.model,
        );

        SELECT 1;
    """
    )
    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    expressions = d.parse(
        """
        MODEL (
            name db.model,
        );

        SELECT 1;

        @noop();
    """
    )
    new_model = load_sql_based_model(
        expressions, path=Path("./examples/sushi/models/test_model.sql")
    )
    assert model.metadata_hash != new_model.metadata_hash
    assert model.data_hash != new_model.data_hash
    assert new_model.is_metadata_only_change(model) == metadata_only

    @macro(metadata_only=metadata_only)  # type: ignore
    def noop(evaluator) -> None:
        print("noop")
        return None

    updated_model = load_sql_based_model(
        expressions, path=Path("./examples/sushi/models/test_model.sql")
    )
    if metadata_only:
        assert "print" not in new_model._additional_metadata[0]
        assert "print" in updated_model._additional_metadata[0]
        assert new_model.data_hash == updated_model.data_hash
        assert new_model.metadata_hash != updated_model.metadata_hash
    else:
        assert "print" not in new_model._data_hash_values[0]
        assert "print" in updated_model._data_hash_values[0]
        assert new_model.data_hash != updated_model.data_hash
        assert new_model.metadata_hash == updated_model.metadata_hash
    assert updated_model.is_metadata_only_change(new_model) == metadata_only


def test_managed_kind_sql():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind MANAGED,
            physical_properties (
                warehouse = small,
                target_lag = '20 minutes',
                refresh_mode = auto
            )
        );

        SELECT a, b
        """
    )

    model = load_sql_based_model(expressions)

    assert model.kind.is_managed

    with pytest.raises(ConfigError, match=r".*must specify the 'target_lag' physical property.*"):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name db.table,
                kind MANAGED,
                dialect snowflake
            );

            SELECT a, b
            """
            )
        ).validate_definition()


def test_managed_kind_python():
    @model("test_managed_python_model", kind="managed", columns={"a": "int"})
    def execute(
        context: ExecutionContext,
        start: datetime,
        end: datetime,
        execution_time: datetime,
        **kwargs: t.Any,
    ) -> pd.DataFrame:
        return pd.DataFrame.from_dict(data={"a": 1}, orient="index")

    with pytest.raises(
        SQLMeshError,
        match=r".*Cannot create Python model.*the 'MANAGED' kind doesn't support Python models",
    ):
        model.get_registry()["test_managed_python_model"].model(
            module_path=Path("."),
            path=Path("."),
        ).validate_definition()


def test_physical_version():
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column a,
                forward_only TRUE,
            ),
            physical_version '1234',
        );

        SELECT a, b
        """
    )

    model = load_sql_based_model(expressions)
    assert model.physical_version == "1234"

    with pytest.raises(
        ConfigError,
        match=r"Pinning a physical version is only supported for forward only models( at.*)?",
    ):
        load_sql_based_model(
            d.parse(
                """
            MODEL (
                name db.table,
                kind INCREMENTAL_BY_TIME_RANGE(
                    time_column a,
                ),
                physical_version '1234',
            );

            SELECT a, b
            """
            )
        ).validate_definition()


def test_trailing_comments():
    expressions = d.parse(
        """
        MODEL (name db.table);

        /* some comment A */

        SELECT 1;
        /* some comment B */
        """
    )
    model = load_sql_based_model(expressions)
    assert not model.render_pre_statements()
    assert not model.render_post_statements()

    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
              batch_size 100,
            )
        );

        /* some comment A */
        """
    )
    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))
    assert not model.render_pre_statements()
    assert not model.render_post_statements()


def test_comments_in_jinja_query():
    expressions = d.parse(
        """
        MODEL (name db.table);

        JINJA_QUERY_BEGIN;
        /* some comment A */

        SELECT 1;
        /* some comment B */

        JINJA_END;
        """
    )
    model = load_sql_based_model(expressions)
    assert model.render_query().sql() == '/* some comment A */ SELECT 1 AS "1"'

    expressions = d.parse(
        """
        MODEL (name db.table);

        JINJA_QUERY_BEGIN;
        /* some comment A */

        SELECT 1;
        SELECT 2;
        /* some comment B */

        JINJA_END;
        """
    )
    model = load_sql_based_model(expressions)
    with pytest.raises(ConfigError, match=r"Too many statements in query.*"):
        model.render_query()


def test_jinja_render_parse_error():
    expressions = d.parse(
        """
        MODEL (name db.test_model);

        JINJA_QUERY_BEGIN;
        {{ unknown_macro() }}
        JINJA_END;
        """
    )

    model = load_sql_based_model(expressions)

    with pytest.raises(ConfigError, match=r"Could not render jinja"):
        model.render_query()


def test_jinja_render_debug_logging(caplog):
    """Test that rendered Jinja expressions are logged for debugging."""
    import logging

    # Set log level to DEBUG to capture debug logs
    caplog.set_level(logging.DEBUG, logger="sqlmesh.core.renderer")

    # Create a model with unparseable Jinja that will be rendered
    expressions = d.parse(
        """
        MODEL (name db.test_model);

        JINJA_QUERY_BEGIN;
        {{ 'SELECT invalid syntax here!' }}
        JINJA_END;
        """
    )

    model = load_sql_based_model(expressions)

    # Attempt to render - this should fail due to invalid SQL syntax
    with pytest.raises(ConfigError, match=r"Could not parse the rendered jinja"):
        model.render_query()

    # Check that the rendered Jinja was logged
    assert any(
        'Rendered Jinja expression for model \'"db"."test_model"\'' in record.message
        and "SELECT invalid syntax here!" in record.message
        for record in caplog.records
    )


def test_staged_file_path():
    expressions = d.parse(
        """
        MODEL (name test, dialect snowflake);

        SELECT * FROM @a.b/c/d.csv(FILE_FORMAT => 'b.ff')
        """
    )
    model = load_sql_based_model(expressions)
    query = model.render_query()
    assert query.sql(dialect="snowflake") == "SELECT * FROM @a.b/c/d.csv (FILE_FORMAT => 'b.ff')"

    expressions = d.parse(
        """
        MODEL (name test, dialect snowflake);

        SELECT
          *
        FROM @variable (FILE_FORMAT => 'foo'), @non_variable (FILE_FORMAT => 'bar')
        LIMIT 100
        """
    )
    model = load_sql_based_model(expressions, variables={"variable": "some_path"})
    query = model.render_query()
    assert (
        query.sql(dialect="snowflake")
        == """SELECT * FROM 'some_path' (FILE_FORMAT => 'foo') AS "SOME_PATH", @non_variable (FILE_FORMAT => 'bar') LIMIT 100"""
    )


def test_cache():
    expressions = d.parse(
        """
        MODEL (name test);

        SELECT 1 x
        FROM y

        """
    )
    model = load_sql_based_model(expressions)
    assert model.depends_on == {'"y"'}
    assert model.copy(update={"depends_on_": {'"z"'}}).depends_on == {'"z"', '"y"'}


def test_snowflake_macro_func_as_table(tmp_path: Path):
    init_example_project(tmp_path, engine_type="duckdb")

    custom_macro_file = tmp_path / "macros/custom_macros.py"
    custom_macro_file.parent.mkdir(parents=True, exist_ok=True)
    custom_macro_file.write_text("""
from sqlmesh import macro

@macro()
def custom_macro(evaluator, arg1, arg2):
    return "SELECT 1 AS c"

@macro()
def get_model_name(evaluator):
    return f"sqlmesh_example.{evaluator._path.stem}"
    """)

    new_snowflake_model_file = tmp_path / "models/new_model.sql"
    new_snowflake_model_file.parent.mkdir(parents=True, exist_ok=True)
    new_snowflake_model_file.write_text("""
MODEL (
  name @get_model_name(),
  dialect snowflake,
);

SELECT * FROM (@custom_macro(@a, @b)) AS q
    """)

    config = Config(
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
        variables={"a": "a", "b": "b"},
    )
    context = Context(paths=tmp_path, config=config)

    query = context.get_model("sqlmesh_example.new_model").render_query()

    assert (
        t.cast(exp.Query, query).sql("snowflake")
        == 'SELECT "Q"."C" AS "C" FROM (SELECT 1 AS "C") AS "Q"'
    )

    context.plan(no_prompts=True, auto_apply=True)


def test_resolve_table(make_snapshot: t.Callable):
    @macro()
    def resolve_parent(evaluator, name):
        return evaluator.resolve_table(name.name)

    for post_statement in (
        "JINJA_STATEMENT_BEGIN; {{ resolve_table('parent') }}; JINJA_END;",
        "@resolve_parent('parent')",
    ):
        expressions = d.parse(
            f"""
            MODEL (name child);

            SELECT c FROM parent;

            {post_statement}
            """
        )
        child = load_sql_based_model(expressions)
        parent = load_sql_based_model(d.parse("MODEL (name parent); SELECT 1 AS c"))

        parent_snapshot = make_snapshot(parent)
        parent_snapshot.categorize_as(SnapshotChangeCategory.BREAKING)
        version = parent_snapshot.version

        post_statements = child.render_post_statements(snapshots={'"parent"': parent_snapshot})

        assert len(post_statements) == 1
        assert post_statements[0].sql() == f'"sqlmesh__default"."parent__{version}"'

    # test with additional nesting level and default catalog
    for post_statement in (
        "JINJA_STATEMENT_BEGIN; {{ resolve_table('schema.parent') }}; JINJA_END;",
        "@resolve_parent('schema.parent')",
    ):
        expressions = d.parse(
            f"""
            MODEL (name schema.child);

            SELECT c FROM schema.parent;

            {post_statement}
            """
        )
        child = load_sql_based_model(expressions, default_catalog="main")
        parent = load_sql_based_model(
            d.parse("MODEL (name schema.parent); SELECT 1 AS c"), default_catalog="main"
        )

        parent_snapshot = make_snapshot(parent)
        parent_snapshot.categorize_as(SnapshotChangeCategory.BREAKING)
        version = parent_snapshot.version

        post_statements = child.render_post_statements(
            snapshots={'"main"."schema"."parent"': parent_snapshot}
        )

        assert len(post_statements) == 1
        assert post_statements[0].sql() == f'"main"."sqlmesh__schema"."schema__parent__{version}"'


def test_cluster_with_complex_expression():
    expressions = d.parse(
        """
        MODEL (
          name test,
          dialect snowflake,
          kind full,
          clustered_by (to_date(cluster_col))
        );

        SELECT
          1 AS c,
          CAST('2020-01-01 12:05:03' AS TIMESTAMPTZ) AS cluster_col
    """
    )

    model = load_sql_based_model(expressions)
    assert [expr.sql("snowflake") for expr in model.clustered_by] == ['(TO_DATE("CLUSTER_COL"))']


def test_parametric_model_kind():
    parsed_definition = d.parse(
        """
        MODEL (
          name db.test_schema.test_model,
          kind @IF(@gateway = 'main', VIEW, FULL)
        );

        SELECT
          1 AS c
        """
    )

    model = load_sql_based_model(parsed_definition, variables={c.GATEWAY: "main"})
    assert isinstance(model.kind, ViewKind)

    model = load_sql_based_model(parsed_definition, variables={c.GATEWAY: "other"})
    assert isinstance(model.kind, FullKind)


def test_fingerprint_signals():
    @signal()
    def test_signal_hash(batch):
        return True

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            signals [
                test_signal_hash(arg = 1),
            ],
        );
        SELECT 1;
        """
    )

    model = load_sql_based_model(expressions, signal_definitions=signal.get_registry())
    metadata_hash = model.metadata_hash
    data_hash = model.data_hash

    def assert_metadata_only():
        model._metadata_hash = None
        model._data_hash = None
        assert model.metadata_hash != metadata_hash
        assert model.data_hash == data_hash

    executable = model.python_env["test_signal_hash"]
    model.python_env["test_signal_hash"].payload = executable.payload.replace("True", "False")
    assert_metadata_only()

    model = load_sql_based_model(expressions, signal_definitions=signal.get_registry())
    model.signals.clear()
    assert_metadata_only()


def test_model_optimize(tmp_path: Path, assert_exp_eq):
    defaults = [
        ModelDefaultsConfig(optimize_query=True).dict(),
        ModelDefaultsConfig(optimize_query=False).dict(),
    ]
    non_optimized_sql = 'SELECT 1 + 2 AS "new_col"'
    optimized_sql = 'SELECT 3 AS "new_col"'

    # Model flag is False, overriding defaults
    disabled_opt = d.parse(
        """
        MODEL (
            name test,
            optimize_query False,
        );

        SELECT 1 + 2 AS new_col
        """
    )

    for default in defaults:
        model = load_sql_based_model(disabled_opt, defaults=default)
        assert_exp_eq(model.render_query(), non_optimized_sql)

    # Model flag is True, overriding defaults
    enabled_opt = d.parse(
        """
        MODEL (
            name test,
            optimize_query True,
        );

        SELECT 1 + 2 AS new_col
        """
    )

    for default in defaults:
        model = load_sql_based_model(enabled_opt, defaults=default)
        assert_exp_eq(model.render_query(), optimized_sql)

    # Model flag is not defined, behavior is set according to the defaults
    none_opt = d.parse(
        """
        MODEL (
            name test,
        );

        SELECT 1 + 2 AS new_col
        """
    )

    assert_exp_eq(load_sql_based_model(none_opt).render_query(), optimized_sql)
    assert_exp_eq(
        load_sql_based_model(none_opt, defaults=defaults[0]).render_query(), optimized_sql
    )
    assert_exp_eq(
        load_sql_based_model(none_opt, defaults=defaults[1]).render_query(), non_optimized_sql
    )

    # Ensure that plan works as expected (optimize_query flag affects the model's data hash)
    for parsed_model in [enabled_opt, disabled_opt, none_opt]:
        context = Context(config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb")))
        context.upsert_model(load_sql_based_model(parsed_model))
        context.plan(auto_apply=True, no_prompts=True)

    # Ensure non-SQLModels raise if optimize_query is not None
    seed_path = tmp_path / "seed.csv"
    model_kind = SeedKind(path=str(seed_path.absolute()))
    with open(seed_path, "w", encoding="utf-8") as fd:
        fd.write(
            """
col_a,col_b,col_c
1,text_a,1.0
2,text_b,2.0"""
        )
    model = create_seed_model("test_db.test_seed_model", model_kind, optimize_query=True)
    context = Context(config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb")))

    with pytest.raises(
        ConfigError,
        match=r"SQLMesh query optimizer can only be enabled for SQL models",
    ):
        context.upsert_model(model)
        context.plan(auto_apply=True, no_prompts=True)

    model = create_seed_model("test_db.test_seed_model", model_kind, optimize_query=False)
    context.upsert_model(model)
    context.plan(auto_apply=True, no_prompts=True)


def test_column_description_metadata_change():
    context = Context(config=Config())

    model = load_sql_based_model(
        d.parse(
            """
        MODEL (
          name db.test_model,
          kind full
        );

        SELECT
          1 AS id /* description */
        """
        ),
        default_catalog=context.default_catalog,
    )

    context.upsert_model(model)
    context.plan(no_prompts=True, auto_apply=True)

    context.upsert_model(
        "db.test_model", query_=ParsableSql(sql="SELECT 1 AS id /* description 2 */")
    )
    plan = context.plan(no_prompts=True, auto_apply=True)

    snapshots = list(plan.snapshots.values())
    assert len(snapshots) == 1

    snapshot = snapshots[0]
    assert len(snapshot.previous_versions) == 1
    assert snapshot.change_category == SnapshotChangeCategory.METADATA


def test_auto_restatement():
    parsed_definition = d.parse(
        """
        MODEL (
          name test_schema.test_model,
          kind INCREMENTAL_BY_TIME_RANGE(
            time_column a,
            auto_restatement_cron '@daily',
          )
        );
        SELECT 1 AS c
        """
    )
    model = load_sql_based_model(parsed_definition)
    assert model.auto_restatement_cron == "@daily"
    assert (
        model.kind.to_expression().sql(pretty=True)
        == """INCREMENTAL_BY_TIME_RANGE (
  time_column ("a", '%Y-%m-%d'),
  partition_by_time_column TRUE,
  forward_only FALSE,
  disable_restatement FALSE,
  on_destructive_change 'ERROR',
  on_additive_change 'ALLOW',
  auto_restatement_cron '@daily'
)"""
    )

    parsed_definition = d.parse(
        """
        MODEL (
          name test_schema.test_model,
          kind INCREMENTAL_BY_TIME_RANGE(
            time_column a,
            auto_restatement_cron '@daily',
            auto_restatement_intervals 1,
          )
        );
        SELECT 1 AS c
        """
    )
    model = load_sql_based_model(parsed_definition)
    assert model.auto_restatement_cron == "@daily"
    assert model.auto_restatement_intervals == 1
    assert (
        model.kind.to_expression().sql(pretty=True)
        == """INCREMENTAL_BY_TIME_RANGE (
  time_column ("a", '%Y-%m-%d'),
  partition_by_time_column TRUE,
  auto_restatement_intervals 1,
  forward_only FALSE,
  disable_restatement FALSE,
  on_destructive_change 'ERROR',
  on_additive_change 'ALLOW',
  auto_restatement_cron '@daily'
)"""
    )

    parsed_definition = d.parse(
        """
        MODEL (
          name test_schema.test_model,
          kind INCREMENTAL_BY_TIME_RANGE(
            time_column a,
            auto_restatement_cron '@invalid'
          )
        );
        SELECT 1 AS c
        """
    )
    with pytest.raises(ValueError, match="Invalid cron expression '@invalid'.*"):
        load_sql_based_model(parsed_definition)


def test_gateway_specific_render(assert_exp_eq) -> None:
    gateways = {
        "main": GatewayConfig(connection=DuckDBConnectionConfig()),
        "duckdb": GatewayConfig(connection=DuckDBConnectionConfig()),
    }
    config = Config(
        gateways=gateways,
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
        default_gateway="main",
    )
    context = Context(config=config)
    assert context.engine_adapter == context.engine_adapters["main"]

    @model(
        name="dummy_model",
        is_sql=True,
        kind="full",
        gateway="duckdb",
        grain='"x"',
    )
    def dummy_model_entry(evaluator: MacroEvaluator) -> exp.Select:
        return exp.select("x").from_(exp.values([("1", 2)], "_v", ["x"]))

    dummy_model = model.get_registry()["dummy_model"].model(module_path=Path("."), path=Path("."))
    context.upsert_model(dummy_model)
    assert isinstance(dummy_model, SqlModel)
    assert dummy_model.gateway == "duckdb"

    assert_exp_eq(
        context.render("dummy_model"),
        """
        SELECT
          "_v"."x" AS "x",
        FROM (VALUES ('1', 2)) AS "_v"("x")
        """,
    )
    assert isinstance(context._get_engine_adapter("duckdb"), DuckDBEngineAdapter)
    assert len(context.engine_adapters) == 2


def test_model_on_virtual_update(make_snapshot: t.Callable):
    # Macro to test resolution within virtual statement
    @macro()
    def resolve_parent_name(evaluator, name):
        return evaluator.resolve_table(name.name)

    dialect = "postgres"
    virtual_update_statements = """
        CREATE OR REPLACE VIEW test_view FROM demo_db.table;
        GRANT SELECT ON VIEW @this_model TO ROLE owner_name;
        JINJA_STATEMENT_BEGIN;
        GRANT SELECT ON VIEW {{this_model}} TO ROLE admin;
        JINJA_END;
        GRANT REFERENCES, SELECT ON FUTURE VIEWS IN DATABASE demo_db TO ROLE owner_name;
        @resolve_parent_name('parent');
        GRANT SELECT ON VIEW demo_db.table /* sqlglot.meta replace=false */ TO ROLE admin;
    """

    expressions = d.parse(
        f"""
        MODEL (
            name demo_db.table,
            owner owner_name,
        );

        SELECT id from parent;

        on_virtual_update_begin;

        {virtual_update_statements}

        on_virtual_update_end;

    """,
        default_dialect=dialect,
    )

    parent_expressions = d.parse(
        """
        MODEL (
            name parent,
        );

        SELECT 1 from id;

        ON_VIRTUAL_UPDATE_BEGIN;
        JINJA_STATEMENT_BEGIN;
        GRANT SELECT ON VIEW {{this_model}} TO ROLE admin;
        JINJA_END;
        ON_VIRTUAL_UPDATE_END;

    """,
        default_dialect=dialect,
    )

    model = load_sql_based_model(expressions, dialect=dialect)
    parent = load_sql_based_model(parent_expressions, dialect=dialect)

    parent_snapshot = make_snapshot(parent)
    parent_snapshot.categorize_as(SnapshotChangeCategory.BREAKING)

    model_snapshot = make_snapshot(model)
    model_snapshot.categorize_as(SnapshotChangeCategory.BREAKING)

    assert model.on_virtual_update == d.parse(virtual_update_statements, default_dialect=dialect)

    assert parent.on_virtual_update == d.parse(
        "JINJA_STATEMENT_BEGIN; GRANT SELECT ON VIEW {{this_model}} TO ROLE admin; JINJA_END;",
        default_dialect=dialect,
    )

    environment_naming_info = EnvironmentNamingInfo(name="dev")
    table_mapping = {model.fqn: "demo_db__dev.table", parent.fqn: "default__dev.parent"}
    snapshots = {
        parent_snapshot.name: parent_snapshot,
        model_snapshot.name: model_snapshot,
    }

    rendered_on_virtual_update = model.render_on_virtual_update(
        snapshots=snapshots,
        table_mapping=table_mapping,
        this_model=model_snapshot.qualified_view_name.table_for_environment(
            environment_naming_info, dialect=dialect
        ),
    )

    assert len(rendered_on_virtual_update) == 6
    assert (
        rendered_on_virtual_update[0].sql()
        == 'CREATE OR REPLACE VIEW "test_view" AS SELECT * FROM "demo_db__dev"."table" AS "table" /* demo_db.table */'
    )

    assert (
        rendered_on_virtual_update[1].sql()
        == 'GRANT SELECT ON VIEW "demo_db__dev"."table" TO ROLE "owner_name"'
    )
    assert (
        rendered_on_virtual_update[3].sql()
        == "GRANT REFERENCES, SELECT ON FUTURE VIEWS IN DATABASE demo_db TO ROLE owner_name"
    )

    assert rendered_on_virtual_update[4].sql() == '"default__dev"."parent"'

    # When replace=false the table should remain as is
    assert (
        rendered_on_virtual_update[5].sql()
        == 'GRANT SELECT ON VIEW "demo_db"."table" /* sqlglot.meta replace=false */ TO ROLE "admin"'
    )

    rendered_parent_on_virtual_update = parent.render_on_virtual_update(
        snapshots=snapshots,
        table_mapping=table_mapping,
        this_model=parent_snapshot.qualified_view_name.table_for_environment(
            environment_naming_info, dialect=dialect
        ),
    )
    assert len(rendered_parent_on_virtual_update) == 1
    assert (
        rendered_parent_on_virtual_update[0].sql()
        == 'GRANT SELECT ON VIEW "default__dev"."parent" TO ROLE "admin"'
    )


def test_python_model_on_virtual_update():
    macros = """
    {% macro index_name(v) %}{{ v }}{% endmacro %}
    """

    jinja_macros = JinjaMacroRegistry()
    jinja_macros.add_macros(MacroExtractor().extract(macros))

    @model(
        "db.test_model",
        kind="full",
        columns={"id": "string", "name": "string"},
        on_virtual_update=[
            "JINJA_STATEMENT_BEGIN;\nCREATE INDEX {{index_name('id_index')}} ON db.test_model(id);\nJINJA_END;",
            parse_one("GRANT SELECT ON VIEW @this_model TO ROLE dev_role;"),
            "GRANT REFERENCES, SELECT ON FUTURE VIEWS IN DATABASE db TO ROLE dev_role;",
        ],
    )
    def model_with_virtual_statements(context, **kwargs):
        return pd.DataFrame(
            [
                {
                    "id": context.var("1"),
                    "name": context.var("var"),
                }
            ]
        )

    python_model = model.get_registry()["db.test_model"].model(
        module_path=Path("."), path=Path("."), dialect="duckdb", jinja_macros=jinja_macros
    )

    assert len(jinja_macros.root_macros) == 1
    assert len(python_model.jinja_macros.root_macros) == 1
    assert "index_name" in python_model.jinja_macros.root_macros
    assert len(python_model.on_virtual_update) == 3

    rendered_statements = python_model._render_statements(
        python_model.on_virtual_update, table_mapping={'"db"."test_model"': "db.test_model"}
    )

    assert (
        rendered_statements[0].sql()
        == 'CREATE INDEX "id_index" ON "db"."test_model" /* db.test_model */("id" NULLS LAST)'
    )
    assert (
        rendered_statements[1].sql()
        == 'GRANT SELECT ON VIEW "db"."test_model" /* db.test_model */ TO ROLE "dev_role"'
    )
    assert (
        rendered_statements[2].sql()
        == "GRANT REFERENCES, SELECT ON FUTURE VIEWS IN DATABASE db TO ROLE dev_role"
    )


def test_compile_time_checks(tmp_path: Path):
    ctx = Context(
        config=Config(
            model_defaults=ModelDefaultsConfig(dialect="duckdb"),
            linter=LinterConfig(
                enabled=True, rules=["ambiguousorinvalidcolumn", "invalidselectstarexpansion"]
            ),
        ),
        paths=tmp_path,
    )

    cfg_err = "Linter detected errors in the code. Please fix them before proceeding."

    # Strict SELECT * expansion
    strict_query = d.parse(
        """
    MODEL (
        name test,
    );

    SELECT * FROM tbl
    """
    )

    with pytest.raises(LinterError, match=cfg_err):
        ctx.upsert_model(load_sql_based_model(strict_query))
        ctx.plan_builder("dev")

    # Strict column resolution
    strict_query = d.parse(
        """
    MODEL (
        name test,
    );

    SELECT foo
    """
    )

    with pytest.raises(LinterError, match=cfg_err):
        ctx.upsert_model(load_sql_based_model(strict_query))
        ctx.plan_builder("dev")


def test_partition_interval_unit():
    expressions = d.parse(
        """
        MODEL (
            name test,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column ds,
            ),
            cron '0 0 1 * *'
        );
        SELECT '2024-01-01' AS ds;
        """
    )
    model = load_sql_based_model(expressions)
    assert model.partition_interval_unit == IntervalUnit.MONTH

    # Partitioning was explicitly set by the user
    expressions = d.parse(
        """
        MODEL (
            name test,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column ds,
            ),
            cron '0 0 1 * *',
            partitioned_by (ds)
        );
        SELECT '2024-01-01' AS ds;
        """
    )
    model = load_sql_based_model(expressions)
    assert model.partition_interval_unit is None


def test_model_blueprinting(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    db_path = str(tmp_path / "db.db")
    db_connection = DuckDBConnectionConfig(database=db_path)

    gateways = {
        "gw1": GatewayConfig(connection=db_connection, variables={"x": 1}),
        "gw2": GatewayConfig(connection=db_connection, variables={"x": 2}),
    }
    config = Config(
        gateways=gateways,
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
    )

    blueprint_sql = tmp_path / "macros" / "identity_macro.py"
    blueprint_sql.parent.mkdir(parents=True, exist_ok=True)
    blueprint_sql.write_text(
        """from sqlmesh import macro

@macro()
def identity(evaluator, value):
    return value
"""
    )
    blueprint_sql = tmp_path / "models" / "blueprint.sql"
    blueprint_sql.parent.mkdir(parents=True, exist_ok=True)
    blueprint_sql.write_text(
        """
        MODEL (
          name @{blueprint}.test_model_sql,
          gateway @identity(@blueprint),
          blueprints ((blueprint := gw1), (blueprint := gw2)),
          kind FULL
        );

        SELECT
          @x AS x
        """
    )
    blueprint_pydf = tmp_path / "models" / "blueprint_df.py"
    blueprint_pydf.parent.mkdir(parents=True, exist_ok=True)
    blueprint_pydf.write_text(
        """
import pandas as pd  # noqa: TID253
from sqlmesh import model


@model(
    "@{blueprint}.test_model_pydf",
    gateway="@blueprint",
    blueprints=[{"blueprint": "gw1"}, {"blueprint": "gw2"}],
    kind="FULL",
    columns={"x": "INT"},
)
def entrypoint(context, *args, **kwargs):
    x_var = context.var("x")
    assert context.blueprint_var("blueprint").startswith("gw")
    return pd.DataFrame({"x": [x_var]})"""
    )
    blueprint_pysql = tmp_path / "models" / "blueprint_sql.py"
    blueprint_pysql.parent.mkdir(parents=True, exist_ok=True)
    blueprint_pysql.write_text(
        """
from sqlmesh import model


@model(
    "@{blueprint}.test_model_pysql",
    gateway="@blueprint",
    blueprints=[{"blueprint": "gw1"}, {"blueprint": "gw2"}],
    kind="FULL",
    is_sql=True,
)
def entrypoint(evaluator):
    x_var = evaluator.var("x")
    assert evaluator.blueprint_var("blueprint", default="").startswith("gw")
    return f'SELECT {x_var} AS x'"""
    )

    context = Context(paths=tmp_path, config=config)
    models = context.models

    # Each of the three model files "expands" into two models
    assert len(models) == 6

    context.plan(no_prompts=True, auto_apply=True, no_diff=True)

    for model_name in ("test_model_sql", "test_model_pydf", "test_model_pysql"):
        for gateway_no in range(1, 3):
            blueprint_value = f"gw{gateway_no}"

            model = models.get(f'"db"."{blueprint_value}"."{model_name}"')
            assert model is not None
            assert "blueprints" not in model.all_fields()

            python_env = model.python_env

            assert python_env.get(c.SQLMESH_VARS) == Executable.value({"x": gateway_no})

            if model_name == "test_model_sql":
                assert c.SQLMESH_BLUEPRINT_VARS not in python_env
            else:
                assert python_env.get(c.SQLMESH_BLUEPRINT_VARS) == Executable.value(
                    {"blueprint": blueprint_value}
                )

            assert context.fetchdf(f"from {model.fqn}").to_dict() == {"x": {0: gateway_no}}

    multi_variable_blueprint_example = tmp_path / "models" / "multi_variable_blueprint_example.sql"
    multi_variable_blueprint_example.parent.mkdir(parents=True, exist_ok=True)
    multi_variable_blueprint_example.write_text(
        """
        MODEL (
          name @{customer}.my_table,
          blueprints (
            (customer := customer1, Customer_Field := 'bar'),
            (customer := customer2, Customer_Field := qux),
          ),
          kind FULL
        );

        SELECT
          @customer_FIELD AS foo,
          @{customer_field} AS foo2,
          @BLUEPRINT_VAR('customer_field') AS foo3,
        FROM @{customer}.my_source
        """
    )

    context = Context(paths=tmp_path, config=config)
    models = context.models

    # The new model expands into another 2 new models
    assert len(models) == 8

    customer1_model = models.get('"db"."customer1"."my_table"')
    assert customer1_model is not None

    customer1_python_env = customer1_model.python_env
    assert customer1_python_env.get(c.SQLMESH_VARS) is None
    assert customer1_python_env.get(c.SQLMESH_BLUEPRINT_VARS) == Executable.value(
        {"customer": SqlValue(sql="customer1"), "customer_field": SqlValue(sql="'bar'")}
    )

    assert t.cast(exp.Expression, customer1_model.render_query()).sql() == (
        """SELECT 'bar' AS "foo", "bar" AS "foo2", 'bar' AS "foo3" FROM "db"."customer1"."my_source" AS "my_source\""""
    )

    customer2_model = models.get('"db"."customer2"."my_table"')
    assert customer2_model is not None

    customer2_python_env = customer2_model.python_env
    assert customer2_python_env.get(c.SQLMESH_VARS) is None
    assert customer2_python_env.get(c.SQLMESH_BLUEPRINT_VARS) == Executable.value(
        {"customer": SqlValue(sql="customer2"), "customer_field": SqlValue(sql="qux")}
    )

    assert t.cast(exp.Expression, customer2_model.render_query()).sql() == (
        '''SELECT "qux" AS "foo", "qux" AS "foo2", "qux" AS "foo3" FROM "db"."customer2"."my_source" AS "my_source"'''
    )


def test_dynamic_blueprinting_using_custom_macro(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    dynamic_template_sql = tmp_path / "models/dynamic_template_custom_macro.sql"
    dynamic_template_sql.parent.mkdir(parents=True, exist_ok=True)
    dynamic_template_sql.write_text(
        """
        MODEL (
          name @customer.some_table,
          kind FULL,
          blueprints @gen_blueprints(),
        );

        SELECT
          @field_a,
          @{field_b} AS field_b
        FROM @customer.some_source

        """
    )

    dynamic_template_py = tmp_path / "models/dynamic_template_custom_macro.py"
    dynamic_template_py.parent.mkdir(parents=True, exist_ok=True)
    dynamic_template_py.write_text(
        """
from sqlmesh import model

@model(
    "@{customer}.some_other_table",
    kind="FULL",
    blueprints="@gen_blueprints()",
    is_sql=True,
)
def entrypoint(evaluator):
    field_a = evaluator.blueprint_var("field_a")
    return f"SELECT {field_a}, @BLUEPRINT_VAR('field_b') AS field_b FROM @customer.some_source"
"""
    )

    gen_blueprints = tmp_path / "macros/gen_blueprints.py"
    gen_blueprints.parent.mkdir(parents=True, exist_ok=True)
    gen_blueprints.write_text(
        """from sqlmesh import macro

@macro()
def gen_blueprints(evaluator):
    return (
        "((customer := customer1, field_a := x, field_b := y),"
        " (customer := customer2, field_a := z, field_b := w))"
    )"""
    )

    ctx = Context(
        config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb")), paths=tmp_path
    )

    assert len(ctx.models) == 4
    assert '"memory"."customer1"."some_table"' in ctx.models
    assert '"memory"."customer2"."some_table"' in ctx.models
    assert '"memory"."customer1"."some_other_table"' in ctx.models
    assert '"memory"."customer2"."some_other_table"' in ctx.models


def test_dynamic_blueprinting_using_each(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    dynamic_template_sql = tmp_path / "models/dynamic_template_each.sql"
    dynamic_template_sql.parent.mkdir(parents=True, exist_ok=True)
    dynamic_template_sql.write_text(
        """
        MODEL (
          name @customer.some_table,
          kind FULL,
          blueprints @EACH(@values, x -> (customer := schema_@x)),
        );

        SELECT
          1 AS c
        """
    )

    dynamic_template_py = tmp_path / "models/dynamic_template_each.py"
    dynamic_template_py.parent.mkdir(parents=True, exist_ok=True)
    dynamic_template_py.write_text(
        """
from sqlmesh import model

@model(
    "@{customer}.some_other_table",
    kind="FULL",
    blueprints="@EACH(@values, x -> (customer := schema_@x))",
    is_sql=True,
)
def entrypoint(evaluator):
    return "SELECT 1 AS c"
"""
    )

    model_defaults = ModelDefaultsConfig(dialect="duckdb")
    variables = {"values": ["customer1", "customer2"]}
    config = Config(model_defaults=model_defaults, variables=variables)
    ctx = Context(config=config, paths=tmp_path)

    assert len(ctx.models) == 4
    assert '"memory"."schema_customer1"."some_table"' in ctx.models
    assert '"memory"."schema_customer2"."some_table"' in ctx.models
    assert '"memory"."schema_customer1"."some_other_table"' in ctx.models
    assert '"memory"."schema_customer2"."some_other_table"' in ctx.models


def test_single_blueprint(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    single_blueprint = tmp_path / "models/single_blueprint.sql"
    single_blueprint.parent.mkdir(parents=True, exist_ok=True)
    single_blueprint.write_text(
        """
        MODEL (
          name @single_blueprint.some_table,
          kind FULL,
          blueprints ((single_blueprint := bar))
        );

        SELECT 1 AS c
        """
    )

    ctx = Context(
        config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb")), paths=tmp_path
    )

    assert len(ctx.models) == 1
    assert '"memory"."bar"."some_table"' in ctx.models


def test_blueprinting_with_quotes(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    template_with_quoted_vars = tmp_path / "models/template_with_quoted_vars.sql"
    template_with_quoted_vars.parent.mkdir(parents=True, exist_ok=True)
    template_with_quoted_vars.write_text(
        """
        MODEL (
          name m.@{bp_var},
          blueprints (
            (bp_var := "a b"),
            (bp_var := 'c d'),
          ),
        );

        SELECT @bp_var AS c1, @{bp_var} AS c2
        """
    )

    ctx = Context(
        config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb")), paths=tmp_path
    )
    assert len(ctx.models) == 2

    m1 = ctx.get_model('m."a b"', raise_if_missing=True)
    m2 = ctx.get_model('m."c d"', raise_if_missing=True)

    assert m1.name == 'm."a b"'
    assert m2.name == 'm."c d"'
    assert t.cast(exp.Query, m1.render_query()).sql() == '''SELECT "a b" AS "c1", "a b" AS "c2"'''
    assert t.cast(exp.Query, m2.render_query()).sql() == '''SELECT 'c d' AS "c1", "c d" AS "c2"'''


def test_blueprint_variable_precedence_sql(tmp_path: Path, assert_exp_eq: t.Callable) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    blueprint_variables = tmp_path / "models/blueprint_variables.sql"
    blueprint_variables.parent.mkdir(parents=True, exist_ok=True)
    blueprint_variables.write_text(
        """
        MODEL (
          name s.@{bp_name},
          blueprints (
            (bp_name := m1, var1 := 'v1', var2 := 'v2'),
            (bp_name := m2, var1 := 'v3'),
          ),
        );

        @DEF(bp_name, override);

        SELECT
          @var1 AS var1_macro_var,
          @{var1} AS var1_identifier,
          @VAR('var1') AS var1_var_macro_func,
          @BLUEPRINT_VAR('var1') AS var1_blueprint_var_macro_func,

          @var2 AS var2_macro_var,
          @{var2} AS var2_identifier,
          @VAR('var2') AS var2_var_macro_func,
          @BLUEPRINT_VAR('var2') AS var2_blueprint_var_macro_func,

          @bp_name AS bp_name_macro_var,
          @{bp_name} AS bp_name_identifier,
          @VAR('bp_name') AS bp_name_var_macro_func,
          @BLUEPRINT_VAR('bp_name') AS bp_name_blueprint_var_macro_func,
        """
    )

    ctx = Context(
        config=Config(
            model_defaults=ModelDefaultsConfig(dialect="duckdb"),
            variables={"var2": "1"},
        ),
        paths=tmp_path,
    )
    assert len(ctx.models) == 2

    m1 = ctx.get_model("s.m1", raise_if_missing=True)
    m2 = ctx.get_model("s.m2", raise_if_missing=True)

    assert_exp_eq(
        m1.render_query(),
        """
        SELECT
          'v1' AS "var1_macro_var",
          "v1" AS "var1_identifier",
          NULL AS "var1_var_macro_func",
          'v1' AS "var1_blueprint_var_macro_func",
          'v2' AS "var2_macro_var",
          "v2" AS "var2_identifier",
          '1' AS "var2_var_macro_func",
          'v2' AS "var2_blueprint_var_macro_func",
          "override" AS "bp_name_macro_var",
          "override" AS "bp_name_identifier",
          NULL AS "bp_name_var_macro_func",
          "m1" AS "bp_name_blueprint_var_macro_func"
        """,
    )
    assert_exp_eq(
        m2.render_query(),
        """
        SELECT
          'v3' AS "var1_macro_var",
          "v3" AS "var1_identifier",
          NULL AS "var1_var_macro_func",
          'v3' AS "var1_blueprint_var_macro_func",
          '1' AS "var2_macro_var",
          "1" AS "var2_identifier",
          '1' AS "var2_var_macro_func",
          NULL AS "var2_blueprint_var_macro_func",
          "override" AS "bp_name_macro_var",
          "override" AS "bp_name_identifier",
          NULL AS "bp_name_var_macro_func",
          "m2" AS "bp_name_blueprint_var_macro_func"
        """,
    )


def test_blueprint_variable_jinja(tmp_path: Path, assert_exp_eq: t.Callable) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    blueprint_variables = tmp_path / "models/blueprint_variables.sql"
    blueprint_variables.parent.mkdir(parents=True, exist_ok=True)
    blueprint_variables.write_text(
        """
        MODEL (
          name s.@{bp_name},
          blueprints (
            (bp_name := m1, var1 := 'v1', var2 := v2),
            (bp_name := m2, var1 := 'v3'),
          ),
        );

        @DEF(bp_name, override);

        JINJA_QUERY_BEGIN;
        SELECT
            {{ blueprint_var('var1') }} AS var1,
            '{{ blueprint_var('var2') }}' AS var2,
            '{{ blueprint_var('var2', 'var2_default') }}' AS var2_default,
            '{{ blueprint_var('bp_name') }}' AS bp_name
        FROM s.{{ blueprint_var('bp_name') }}_source;
        JINJA_END;
        """
    )
    ctx = Context(
        config=Config(
            model_defaults=ModelDefaultsConfig(dialect="duckdb"),
            variables={"var2": "1"},
        ),
        paths=tmp_path,
    )
    assert len(ctx.models) == 2

    m1 = ctx.get_model("s.m1", raise_if_missing=True)
    m2 = ctx.get_model("s.m2", raise_if_missing=True)

    assert_exp_eq(
        m1.render_query(),
        """SELECT 'v1' AS "var1", 'v2' AS "var2", 'v2' AS "var2_default", 'm1' AS "bp_name" FROM "memory"."s"."m1_source" AS "m1_source" """,
    )
    assert_exp_eq(
        m2.render_query(),
        """SELECT 'v3' AS "var1", 'None' AS "var2", 'var2_default' AS "var2_default", 'm2' AS "bp_name" FROM "memory"."s"."m2_source" AS "m2_source" """,
    )


def test_blueprint_variable_precedence_python(tmp_path: Path, mocker: MockerFixture) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    blueprint_variables = tmp_path / "models/blueprint_variables.py"
    blueprint_variables.parent.mkdir(parents=True, exist_ok=True)
    blueprint_variables.write_text(
        """
import pandas as pd  # noqa: TID253
from sqlglot import exp
from sqlmesh import model


@model(
    "s.@{bp_name}",
    blueprints=[{"bp_name": "m", "var1": exp.to_column("v1"), "var2": 1}],
    kind="FULL",
    columns={"x": "INT"},
)
def entrypoint(context, *args, **kwargs):
    assert "bp_name" not in kwargs
    assert "var1" not in kwargs
    assert kwargs.get("var2") == "1"

    assert context.var("bp_name") is None
    assert context.var("var1") is None
    assert context.var("var2") == "1"

    assert context.blueprint_var("bp_name") == "m"
    assert context.blueprint_var("var1") == exp.to_column("v1")
    assert context.blueprint_var("var2") == 1

    return pd.DataFrame({"x": [1]})
        """
    )

    ctx = Context(
        config=Config(
            model_defaults=ModelDefaultsConfig(dialect="duckdb"),
            variables={"var2": "1"},
        ),
        paths=tmp_path,
    )
    assert len(ctx.models) == 1

    m = ctx.get_model("s.m", raise_if_missing=True)
    context = ExecutionContext(mocker.Mock(), {}, None, None)

    assert t.cast(pd.DataFrame, list(m.render(context=context))[0]).to_dict() == {"x": {0: 1}}


def test_python_model_depends_on_blueprints(tmp_path: Path) -> None:
    sql_model = tmp_path / "models" / "base_blueprints.sql"
    sql_model.parent.mkdir(parents=True, exist_ok=True)
    sql_model.write_text(
        """
        MODEL (
          name test_schema1.@{model_name},
          blueprints ((model_name := foo), (model_name := bar)),
          kind FULL
        );

        SELECT 1 AS id
        """
    )

    py_model = tmp_path / "models" / "depends_on_with_blueprint_vars.py"
    py_model.parent.mkdir(parents=True, exist_ok=True)
    py_model.write_text(
        """
import pandas as pd  # noqa: TID253
from sqlmesh import model

@model(
    "test_schema2.@model_name",
    columns={
        "id": "int",
    },
    blueprints=[
        {"model_name": "foo"},
        {"model_name": "bar"},
    ],
    depends_on=["test_schema1.@{model_name}"],
)
def entrypoint(context, *args, **kwargs):
    table = context.resolve_table(f"test_schema1.{context.blueprint_var('model_name')}")
    return context.fetchdf(f"SELECT * FROM {table}")"""
    )

    ctx = Context(
        config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb")),
        paths=tmp_path,
    )
    assert len(ctx.models) == 4

    ctx.plan(no_prompts=True, auto_apply=True)
    assert ctx.fetchdf("SELECT * FROM test_schema2.foo").to_dict() == {"id": {0: 1}}


@time_machine.travel("2020-01-01 00:00:00 UTC")
def test_dynamic_date_spine_model(assert_exp_eq):
    @macro()
    def get_current_date(evaluator):
        from sqlmesh.utils.date import now

        return f"'{now().date()}'"

    expressions = d.parse(
        """
        MODEL (name test_model, dialect duckdb);

        @DEF(curr_date, @get_current_date());

        WITH discount_promotion_dates AS (
          @date_spine('day', @curr_date::date - 90, @curr_date::date)
        )

        SELECT * FROM discount_promotion_dates
        """
    )
    model = load_sql_based_model(expressions)
    assert_exp_eq(
        model.render_query(),
        """
        WITH "discount_promotion_dates" AS (
          SELECT
            "_exploded"."date_day" AS "date_day"
          FROM UNNEST(CAST(GENERATE_SERIES(CAST('2020-01-01' AS DATE) - 90, CAST('2020-01-01' AS DATE), INTERVAL '1' DAY) AS DATE[])) AS "_exploded"("date_day")
        )
        SELECT
          "discount_promotion_dates"."date_day" AS "date_day"
        FROM "discount_promotion_dates" AS "discount_promotion_dates"
        """,
    )


def test_seed_dont_coerce_na_into_null(tmp_path):
    model_csv_path = (tmp_path / "model.csv").absolute()

    with open(model_csv_path, "w", encoding="utf-8") as fd:
        fd.write("code\nNA")

    expressions = d.parse(
        f"""
        MODEL (
            name db.seed,
            kind SEED (
              path '{str(model_csv_path)}',
              csv_settings (
                -- override NaN handling, such that no value can be coerced into NaN
                keep_default_na = false,
                na_values = (),
              ),
            ),
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))

    assert isinstance(model.kind, SeedKind)
    assert model.seed is not None
    assert len(model.seed.content) > 0
    assert next(model.render(context=None)).to_dict() == {"code": {0: "NA"}}


def test_seed_coerce_datetime(tmp_path):
    model_csv_path = (tmp_path / "model.csv").absolute()

    with open(model_csv_path, "w", encoding="utf-8") as fd:
        fd.write("bad_datetime\n9999-12-31 23:59:59")

    expressions = d.parse(
        f"""
        MODEL (
            name db.seed,
            kind SEED (
              path '{str(model_csv_path)}',
            ),
            columns (
              bad_datetime datetime,
            ),
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))
    df = next(model.render(context=None))
    assert df["bad_datetime"].iloc[0] == "9999-12-31 23:59:59"


def test_seed_invalid_date_column(tmp_path):
    model_csv_path = (tmp_path / "model.csv").absolute()

    with open(model_csv_path, "w", encoding="utf-8") as fd:
        fd.write("bad_date\n9999-12-31\n2025-01-01\n1000-01-01")

    expressions = d.parse(
        f"""
        MODEL (
            name db.seed,
            kind SEED (
              path '{str(model_csv_path)}',
            ),
            columns (
              bad_date date,
            ),
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))
    df = next(model.render(context=None))
    # The conversion to date should not raise an error
    assert df["bad_date"].to_list() == ["9999-12-31", "2025-01-01", "1000-01-01"]


def test_seed_missing_columns(tmp_path):
    model_csv_path = (tmp_path / "model.csv").absolute()

    with open(model_csv_path, "w", encoding="utf-8") as fd:
        fd.write("key,value\n1,2\n3,4")

    expressions = d.parse(
        f"""
        MODEL (
            name db.seed,
            kind SEED (
              path '{str(model_csv_path)}',
            ),
            columns (
              key int,
              value int,
              missing_column int,
            ),
        );
    """
    )

    model = load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))
    with pytest.raises(
        ConfigError, match="Seed model 'db.seed' has missing columns: {'missing_column'}.*"
    ):
        next(model.render(context=None))


def test_missing_column_data_in_columns_key():
    expressions = d.parse(
        """
        MODEL (
            name db.seed,
            kind SEED (
              path '../seeds/waiter_names.csv',
            ),
            columns (
              culprit, other_column double,
            )
        );
    """
    )
    with pytest.raises(ConfigError, match="Missing data type for column 'culprit'."):
        load_sql_based_model(expressions, path=Path("./examples/sushi/models/test_model.sql"))


def test_ignored_rules_serialization():
    expressions = d.parse(
        """
        MODEL(
            name test_model,
            ignored_rules ['foo', 'bar']
        );

        SELECT * FROM tbl;
    """,
        default_dialect="bigquery",
    )

    model = load_sql_based_model(expressions)

    model_json = model.json()
    model_json_parsed = json.loads(model_json)

    assert "ignored_rules" not in model_json_parsed
    assert "ignored_rules_" not in model_json_parsed

    deserialized_model = SqlModel.parse_raw(model_json)
    assert deserialized_model.dict() == model.dict()


def test_data_hash_unchanged_when_column_type_uses_default_dialect():
    model = create_sql_model(
        "foo",
        parse_one("SELECT * FROM bla"),
        columns={"a": exp.DataType.build("int")},
        dialect="bigquery",
    )

    deserialized_model = SqlModel.parse_raw(model.json())

    assert model.columns_to_types_ == {"a": exp.DataType.build("int")}
    assert deserialized_model.columns_to_types_ == {"a": exp.DataType.build("bigint")}

    # int == int64 in bigquery
    assert model.data_hash == deserialized_model.data_hash


def test_transitive_dependency_of_metadata_only_object_is_metadata_only(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    test_model = tmp_path / "models/test_model.sql"
    test_model.parent.mkdir(parents=True, exist_ok=True)
    test_model.write_text("MODEL (name test_model, kind FULL); @metadata_macro(); SELECT 1 AS c")

    metadata_macro_code = """
from sqlglot import parse_one
from sqlmesh import macro

def get_parsed_query():
    return parse_one("SELECT 1")

@macro(metadata_only=True)
def metadata_macro(evaluator):
    if evaluator.runtime_stage == "evaluating":
        evaluator.engine_adapter.execute({query})"""

    metadata_macro = tmp_path / "macros/metadata_macro.py"
    metadata_macro.parent.mkdir(parents=True, exist_ok=True)
    metadata_macro.write_text(metadata_macro_code.format(query='"SELECT 1"'))

    ctx = Context(
        config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb")),
        paths=tmp_path,
    )

    model = ctx.get_model("test_model")
    python_env = model.python_env

    empty_executable = Executable(payload="")

    assert len(python_env) == 1
    assert (python_env.get("metadata_macro") or empty_executable).is_metadata

    ctx.plan(no_prompts=True, auto_apply=True)

    # This should make `parse_one` a dependency and so it'll be added in the python env
    metadata_macro.write_text(metadata_macro_code.format(query='parse_one("SELECT 1")'))

    ctx.load()
    model = ctx.get_model("test_model")
    python_env = model.python_env

    assert len(python_env) == 2
    assert (python_env.get("metadata_macro") or empty_executable).is_metadata
    assert (python_env.get("parse_one") or empty_executable).is_metadata

    plan = ctx.plan(no_prompts=True, auto_apply=True)
    ctx_diff = plan.context_diff

    assert len(ctx_diff.modified_snapshots) == 1

    new_snapshot, _ = ctx_diff.modified_snapshots['"test_model"']
    assert new_snapshot.change_category == SnapshotChangeCategory.METADATA

    # This should make `get_parsed_query` a dependency and so it'll be added in
    # the python env, carrying `parse_one` with it as another transitive dependency
    metadata_macro.write_text(metadata_macro_code.format(query="get_parsed_query()"))

    ctx.load()
    model = ctx.get_model("test_model")
    python_env = model.python_env

    assert len(python_env) == 3
    assert (python_env.get("metadata_macro") or empty_executable).is_metadata
    assert (python_env.get("get_parsed_query") or empty_executable).is_metadata
    assert (python_env.get("parse_one") or empty_executable).is_metadata

    plan = ctx.plan(no_prompts=True, auto_apply=True)
    ctx_diff = plan.context_diff

    assert len(ctx_diff.modified_snapshots) == 1

    new_snapshot, _ = ctx_diff.modified_snapshots['"test_model"']
    assert new_snapshot.change_category == SnapshotChangeCategory.METADATA


def test_vars_are_taken_into_account_when_propagating_metadata_status(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    test_model = tmp_path / "models/test_model.sql"
    test_model.parent.mkdir(parents=True, exist_ok=True)
    test_model.write_text(
        "MODEL (name test_model, kind FULL, blueprints ((v4 := 4, v5 := 5)));"
        "@m1_metadata_references_v1();"  # metadata macro, references v1 internally => v1 metadata
        "@m2_metadata_does_not_reference_var(@v2, @v3);"  # metadata macro => v2 metadata, v3 metadata
        "@m3_non_metadata_references_v4(@v3);"  # non-metadata macro, references v4 => v3, v4 are not metadata
        "SELECT 1 AS c;"
        "@m2_metadata_does_not_reference_var(@v6);"  # metadata macro => v6 is metadata
        "@m4_non_metadata_references_v6();"  # non-metadata macro, references v6 => v6 is not metadata
        "ON_VIRTUAL_UPDATE_BEGIN;"
        "@m3_non_metadata_references_v4(@v5);"  # non-metadata macro, metadata expression => v5 metadata
        "ON_VIRTUAL_UPDATE_END;"
    )

    macro_code = """
from sqlmesh import macro

@macro(metadata_only=True)
def m1_metadata_references_v1(evaluator):
    evaluator.var("v1")
    return None

@macro(metadata_only=True)
def m2_metadata_does_not_reference_var(evaluator, *args):
    return None

@macro()
def m3_non_metadata_references_v4(evaluator, *args):
    evaluator.var("v4")
    return None

@macro()
def m4_non_metadata_references_v6(evaluator):
    evaluator.var("v6")
    return None"""

    test_macros = tmp_path / "macros/test_macros.py"
    test_macros.parent.mkdir(parents=True, exist_ok=True)
    test_macros.write_text(macro_code)

    ctx = Context(
        config=Config(
            model_defaults=ModelDefaultsConfig(dialect="duckdb"),
            variables={"v1": 1, "v2": 2, "v3": 3, "v6": 6},
        ),
        paths=tmp_path,
    )
    model = ctx.get_model("test_model")

    python_env = model.python_env

    assert len(python_env) == 8
    assert "m1_metadata_references_v1" in python_env
    assert "m2_metadata_does_not_reference_var" in python_env
    assert "m3_non_metadata_references_v4" in python_env
    assert "m4_non_metadata_references_v6" in python_env

    variables = python_env.get(c.SQLMESH_VARS)
    metadata_variables = python_env.get(c.SQLMESH_VARS_METADATA)

    assert variables == Executable.value({"v3": 3, "v6": 6})
    assert metadata_variables == Executable.value({"v1": 1, "v2": 2}, is_metadata=True)

    blueprint_variables = python_env.get(c.SQLMESH_BLUEPRINT_VARS)
    blueprint_metadata_variables = python_env.get(c.SQLMESH_BLUEPRINT_VARS_METADATA)

    assert blueprint_variables == Executable.value({"v4": SqlValue(sql="4")})
    assert blueprint_metadata_variables == Executable.value(
        {"v5": SqlValue(sql="5")}, is_metadata=True
    )

    macro_evaluator = MacroEvaluator(python_env=python_env)

    assert macro_evaluator.locals == {
        "runtime_stage": "loading",
        "default_catalog": None,
        c.SQLMESH_VARS: {"v3": 3, "v6": 6},
        c.SQLMESH_VARS_METADATA: {"v1": 1, "v2": 2},
        c.SQLMESH_BLUEPRINT_VARS: {"v4": exp.Literal.number("4")},
        c.SQLMESH_BLUEPRINT_VARS_METADATA: {"v5": exp.Literal.number("5")},
    }
    assert macro_evaluator.var("v1") == 1
    assert macro_evaluator.var("v2") == 2
    assert macro_evaluator.var("v3") == 3
    assert macro_evaluator.var("v6") == 6
    assert macro_evaluator.blueprint_var("v4") == exp.Literal.number("4")
    assert macro_evaluator.blueprint_var("v5") == exp.Literal.number("5")

    query_with_vars = macro_evaluator.transform(
        parse_one("SELECT " + ", ".join(f"@v{var}, @VAR('v{var}')" for var in [1, 2, 3, 6]))
    )
    assert t.cast(exp.Expression, query_with_vars).sql() == "SELECT 1, 1, 2, 2, 3, 3, 6, 6"

    query_with_blueprint_vars = macro_evaluator.transform(
        parse_one("SELECT " + ", ".join(f"@v{var}, @BLUEPRINT_VAR('v{var}')" for var in [4, 5]))
    )
    assert t.cast(exp.Expression, query_with_blueprint_vars).sql() == "SELECT 4, 4, 5, 5"


def test_variable_mentioned_in_both_metadata_and_non_metadata_macro(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    test_model = tmp_path / "models/test_model.sql"
    test_model.parent.mkdir(parents=True, exist_ok=True)
    test_model.write_text(
        "MODEL (name test_model, kind FULL); @m1_references_v_metadata(); SELECT @m2_references_v_non_metadata() AS c;"
    )

    macro_code = """
from sqlmesh import macro

@macro(metadata_only=True)
def m1_references_v_metadata(evaluator):
    evaluator.var("v")
    return None

@macro()
def m2_references_v_non_metadata(evaluator):
    evaluator.var("v")
    return None"""

    test_macros = tmp_path / "macros/test_macros.py"
    test_macros.parent.mkdir(parents=True, exist_ok=True)
    test_macros.write_text(macro_code)

    ctx = Context(
        config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb"), variables={"v": 1}),
        paths=tmp_path,
    )
    model = ctx.get_model("test_model")

    python_env = model.python_env

    assert len(python_env) == 3
    assert set(python_env) > {"m1_references_v_metadata", "m2_references_v_non_metadata"}
    assert python_env.get(c.SQLMESH_VARS) == Executable.value({"v": 1})


def test_only_top_level_macro_func_impacts_var_descendant_metadata_status(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    test_model = tmp_path / "models/test_model.sql"
    test_model.parent.mkdir(parents=True, exist_ok=True)
    test_model.write_text(
        "MODEL (name test_model, kind FULL); @m1_metadata(@m2_non_metadata(@v)); SELECT 1 AS c;"
    )

    macro_code = """
from sqlmesh import macro

@macro(metadata_only=True)
def m1_metadata(evaluator, *args):
    return None

@macro()
def m2_non_metadata(evaluator, *args):
    return None"""

    test_macros = tmp_path / "macros/test_macros.py"
    test_macros.parent.mkdir(parents=True, exist_ok=True)
    test_macros.write_text(macro_code)

    ctx = Context(
        config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb"), variables={"v": 1}),
        paths=tmp_path,
    )
    model = ctx.get_model("test_model")

    python_env = model.python_env

    assert len(python_env) == 3
    assert set(python_env) > {"m1_metadata", "m2_non_metadata"}
    assert python_env.get(c.SQLMESH_VARS_METADATA) == Executable.value({"v": 1}, is_metadata=True)


def test_non_metadata_object_takes_precedence_over_metadata_only_object(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    test_model = tmp_path / "models/test_model.sql"
    test_model.parent.mkdir(parents=True, exist_ok=True)
    test_model.write_text("MODEL (name test_model, kind FULL); @m1(); @m2(); SELECT 1 AS c")

    macro_code = """
from sqlglot import parse_one
from sqlmesh import macro

def common_dep():
    pass

def m1_dep():
    pass

@macro(metadata_only=True)
def m1(evaluator):
    m1_dep()
    common_dep()
    if evaluator.runtime_stage == "evaluating":
        evaluator.engine_adapter.execute(parse_one("SELECT 1"))

@macro()
def m2(evaluator):
    common_dep()
    if evaluator.runtime_stage == "evaluating":
        evaluator.engine_adapter.execute(parse_one("SELECT 1"))"""

    test_macros = tmp_path / "macros/test_macros.py"
    test_macros.parent.mkdir(parents=True, exist_ok=True)
    test_macros.write_text(macro_code)

    ctx = Context(
        config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb")),
        paths=tmp_path,
    )
    model = ctx.get_model("test_model")
    empty_executable = Executable(payload="")

    python_env = model.python_env

    # Both `m1` and `m2` refer to `parse_one`, so for `m1` it would be a transitive metadata-only
    # object, but since the python env is a flat namespace and `parse_one` is also a dependency
    # of `m2`, it needs to be treated as non-metadata.
    assert len(python_env) == 5
    assert (python_env.get("m1") or empty_executable).is_metadata
    assert (python_env.get("m1_dep") or empty_executable).is_metadata
    assert not (python_env.get("m2") or empty_executable).is_metadata
    assert not (python_env.get("parse_one") or empty_executable).is_metadata
    assert not (python_env.get("common_dep") or empty_executable).is_metadata


def test_macros_referenced_in_metadata_statements_and_properties_are_metadata_only(
    tmp_path: Path,
) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    test_model = tmp_path / "models/test_model.sql"
    test_model.parent.mkdir(parents=True, exist_ok=True)
    test_model.write_text(
        """
        MODEL (
          name test_model,
          kind FULL,
          signals (
            test_signal_always_true(arg1 := @m1(), arg2 := @non_metadata_macro())
          ),
          audits (
            non_zero_c1,
            unique_values(columns := @m2()),
          ),
          physical_properties (
            random_prop = @random_prop_macro()
          ),
        );

        SELECT
          1 AS c1,
          @zero_alt() AS c2,
          @zero_metadata() AS c3,
          @non_metadata_macro() AS c4;


        ON_VIRTUAL_UPDATE_BEGIN;

        @bla();
        @random_prop_macro();

        ON_VIRTUAL_UPDATE_END;

        """
    )

    macro_code = """
from sqlglot import exp
from sqlmesh import macro

def baz():
    pass

def bob():
    pass

@macro()
def bla():
    bob()

@macro()
def m1(evaluator):
    baz()
    return 1

@macro()
def m2(evaluator):
    return exp.column("c")

@macro()
def zero(evaluator):
    return 0

@macro()
def zero_alt(evaluator):
    return 0

@macro(metadata_only=True)
def zero_metadata(evaluator):
    return 0

@macro()
def non_metadata_macro(evaluator):
    return 1

@macro()
def random_prop_macro(evaluator):
    return 1"""

    test_macros = tmp_path / "macros/test_macros.py"
    test_macros.parent.mkdir(parents=True, exist_ok=True)
    test_macros.write_text(macro_code)

    signal_code = """
import typing as t

from sqlmesh import signal

def bar():
    pass

@signal()
def test_signal_always_true(batch, arg1, arg2):
    bar()
    return True"""

    test_signals = tmp_path / "signals/test_signals.py"
    test_signals.parent.mkdir(parents=True, exist_ok=True)
    test_signals.write_text(signal_code)

    audit_code = """
    AUDIT (
      name non_zero_c1,
    );

    SELECT
      *
    FROM @this_model
    WHERE
      c = @zero() OR c = @zero_alt() OR c = @zero_metadata();
    """

    test_audits = tmp_path / "audits/non_zero_c1.sql"
    test_audits.parent.mkdir(parents=True, exist_ok=True)
    test_audits.write_text(audit_code)

    ctx = Context(
        config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb")),
        paths=tmp_path,
    )
    model = ctx.get_model("test_model")
    empty_executable = Executable(payload="")

    python_env = model.python_env

    assert len(python_env) == 13
    assert (python_env.get("test_signal_always_true") or empty_executable).is_metadata
    assert (python_env.get("bar") or empty_executable).is_metadata
    assert (python_env.get("m1") or empty_executable).is_metadata
    assert (python_env.get("baz") or empty_executable).is_metadata
    assert (python_env.get("m2") or empty_executable).is_metadata
    assert (python_env.get("exp") or empty_executable).is_metadata
    assert (python_env.get("zero") or empty_executable).is_metadata
    assert (python_env.get("zero_metadata") or empty_executable).is_metadata
    assert (python_env.get("bla") or empty_executable).is_metadata
    assert (python_env.get("bob") or empty_executable).is_metadata

    # non_metadata_macro is referenced in the signal, which makes that reference "metadata only",
    # but it's also referenced in the model's query and the macro itself is not "metadata only",
    # so the corresponding executable needs to be included in the data hash calculation. The same
    # is true for zero_alt, which is referenced in the non_zero_c1 audit.
    assert not (python_env.get("zero_alt") or empty_executable).is_metadata
    assert not (python_env.get("non_metadata_macro") or empty_executable).is_metadata
    assert not (python_env.get("random_prop_macro") or empty_executable).is_metadata


def test_scd_type_2_full_history_restatement():
    assert ModelKindName.SCD_TYPE_2.full_history_restatement_only is True
    assert ModelKindName.SCD_TYPE_2_BY_TIME.full_history_restatement_only is True
    assert ModelKindName.SCD_TYPE_2_BY_COLUMN.full_history_restatement_only is True
    assert ModelKindName.INCREMENTAL_BY_TIME_RANGE.full_history_restatement_only is False


def test_python_model_boolean_values():
    @model(
        "db.test_model",
        kind=dict(
            name=ModelKindName.SCD_TYPE_2_BY_TIME,
            unique_key=["id"],
            disable_restatement=False,
        ),
        columns={"id": "string", "name": "string"},
        optimize_query=False,
    )
    def test_model(context, **kwargs):
        return pd.DataFrame([{"id": context.var("1")}])

    python_model = model.get_registry()["db.test_model"].model(
        module_path=Path("."), path=Path("."), dialect="duckdb"
    )

    assert python_model.kind.disable_restatement is False
    assert python_model.optimize_query is False


def test_var_in_def(assert_exp_eq):
    expressions = d.parse(
        """
        MODEL (
            name db.table,
            kind INCREMENTAL_BY_TIME_RANGE(
                time_column ds
            ),
        );

        @DEF(var, @start_ds);

        SELECT @var AS ds
    """
    )

    model = load_sql_based_model(expressions)

    assert_exp_eq(
        model.render_query(),
        """
        SELECT '1970-01-01' AS "ds"
        """,
    )


def test_formatting_flag_serde():
    expressions = d.parse(
        """
        MODEL(
            name test_model,
            formatting False,
        );
        SELECT * FROM tbl;
    """,
        default_dialect="duckdb",
    )

    model = load_sql_based_model(expressions)
    assert model.render_definition()[0].sql() == "MODEL (\nname test_model,\nformatting False\n)"

    model_json = model.json()
    assert "formatting" not in json.loads(model_json)

    deserialized_model = SqlModel.parse_raw(model_json)
    assert deserialized_model.dict() == model.dict()


def test_call_python_macro_from_jinja():
    def noop() -> None:
        print("noop")

    @macro()
    def test_runtime_stage(evaluator):
        noop()
        return evaluator.runtime_stage

    expressions = d.parse(
        """
        MODEL (
            name db.table,
            dialect spark,
            owner owner_name,
        );

        JINJA_QUERY_BEGIN;
        SELECT '{{ test_runtime_stage() }}' AS a, '{{ test_runtime_stage_jinja('bla') }}' AS b;
        JINJA_END;
    """
    )

    jinja_macros = JinjaMacroRegistry(
        root_macros={
            "test_runtime_stage_jinja": MacroInfo(
                definition="{% macro test_runtime_stage_jinja(value) %}{{ test_runtime_stage() }}_{{ value }}{% endmacro %}",
                depends_on=[],
            )
        }
    )

    model = load_sql_based_model(expressions, jinja_macros=jinja_macros)
    assert model.render_query().sql() == "SELECT 'loading' AS a, 'loading_bla' AS b"
    assert set(model.python_env) == {"noop", "test_runtime_stage"}


def test_python_env_references_are_unequal_but_point_to_same_definition(tmp_path: Path) -> None:
    # This tests for regressions against an edge case bug which was due to reloading modules
    # in sqlmesh.utils.metaprogramming.import_python_file. Depending on the module loading
    # order, we could get a "duplicate symbol in python env" error, even though the references
    # essentially pointed to the same definition (e.g. function or class).
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    db_path = str(tmp_path / "db.db")
    db_connection = DuckDBConnectionConfig(database=db_path)

    config = Config(
        gateways={"duckdb": GatewayConfig(connection=db_connection)},
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
    )

    file_a = tmp_path / "macros" / "a.py"
    file_b = tmp_path / "macros" / "b.py"
    file_c = tmp_path / "macros" / "c.py"

    file_a.write_text(
        """from macros.c import target

def f1():
    target()
"""
    )
    file_b.write_text(
        """from sqlmesh import macro

from macros.a import f1
from macros.c import target

@macro()
def first_macro(evaluator):
    f1()

@macro()
def second_macro(evaluator):
    target()
"""
    )
    file_c.write_text(
        """def target():
    pass
"""
    )

    model_file = tmp_path / "models" / "model.sql"
    model_file.write_text("MODEL (name a); @first_macro(); @second_macro(); SELECT 1 AS c")

    ctx = Context(paths=tmp_path, config=config, load=False)
    loader = ctx._loaders[0]

    original_glob_paths = loader._glob_paths

    def _patched_glob_paths(path, *args, **kwargs):
        if path == tmp_path / "macros":
            yield from [file_a, file_c, file_b]
        else:
            yield from original_glob_paths(path, *args, **kwargs)

    # We force the import order to be a.py -> c.py -> b.py:
    #
    # 1. a.py is loaded, so "macros", "macros.a" and "macros.c" are loaded in sys.modules
    # 2. c.py is loaded, so "macros" and "macros.c" are reloaded in sys.modules
    # 3. b.py is loaded, so "macros" is reloaded and "macros.b" is loaded in sys.modules
    #
    # (1) => id(sys.modules["macros.a"].target) == id(sys.modules["macros.c"].target) == X
    # (2) => id(sys.modules["macros.c"].target) == Y != X == id(sys.modules["macros.a"].target)
    # (3) => affects neither sys.modules["macros.a"] nor sys.modules["macros.c"], just loads the macros
    #
    # At this point we have two different function instances, one in sys.modules["macros.a"] and one
    # in sys.modules["macros.c"], which encapsulate the same definition (source code). This used to
    # lead to a crash, because we prohibit unequal objects with the same name in the python env.
    with patch.object(loader, "_glob_paths", side_effect=_patched_glob_paths):
        ctx.load()

    model = ctx.models['"a"']
    python_env = model.python_env

    assert len(python_env) == 4
    assert python_env.get("target") == Executable(
        payload="def target():\n    pass", name="target", path="macros/c.py"
    )


def test_unequal_duplicate_python_env_references_are_prohibited(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    db_path = str(tmp_path / "db.db")
    db_connection = DuckDBConnectionConfig(database=db_path)
    config = Config(
        gateways={"duckdb": GatewayConfig(connection=db_connection)},
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
    )

    file_a = tmp_path / "macros" / "unimportant_macro.py"
    file_b = tmp_path / "macros" / "just_f.py"

    file_a.write_text(
        """from sqlmesh import macro
from macros.just_f import f

a = False

@macro()
def unimportant_macro(evaluator):
    print(a)
    f()
    return 1
"""
    )
    file_b.write_text(
        """a = 0

def f():
    print(a)
"""
    )

    model_file = tmp_path / "models" / "model.sql"
    model_file.write_text("MODEL (name m); SELECT @unimportant_macro() AS unimportant_macro")

    with pytest.raises(SQLMeshError, match=r"duplicate definitions found"):
        Context(paths=tmp_path, config=config)


def test_semicolon_is_metadata_only_change(tmp_path, assert_exp_eq):
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    db_connection = DuckDBConnectionConfig(database=str(tmp_path / "db.db"))
    config = Config(
        gateways={"duckdb": GatewayConfig(connection=db_connection)},
        model_defaults=ModelDefaultsConfig(dialect="duckdb"),
    )

    model_file = tmp_path / "models" / "model_with_semicolon.sql"
    model_file.write_text(
        """
        MODEL (
          name sqlmesh_example.incremental_model_with_semicolon,
          kind INCREMENTAL_BY_TIME_RANGE (
            time_column event_date
          ),
          start '2020-01-01',
          cron '@daily',
          grain (id, event_date)
        );

        SELECT
          1 AS id,
          1 AS item_id,
          CAST('2020-01-01' AS DATE) AS event_date
        ;

        --Just a comment
        """
    )

    ctx = Context(paths=tmp_path, config=config)
    model = ctx.get_model("sqlmesh_example.incremental_model_with_semicolon")

    assert not model.pre_statements
    assert not model.post_statements

    assert_exp_eq(
        model.render_query(),
        'SELECT 1 AS "id", 1 AS "item_id", CAST(\'2020-01-01\' AS DATE) AS "event_date"',
    )
    ctx.format()

    assert (
        model_file.read_text()
        == """MODEL (
  name sqlmesh_example.incremental_model_with_semicolon,
  kind INCREMENTAL_BY_TIME_RANGE (
    time_column event_date
  ),
  start '2020-01-01',
  cron '@daily',
  grain (id, event_date)
);

SELECT
  1 AS id,
  1 AS item_id,
  '2020-01-01'::DATE AS event_date;

/* Just a comment */"""
    )

    ctx.plan(no_prompts=True, auto_apply=True)

    model_file = tmp_path / "models" / "model_with_semicolon.sql"
    model_file.write_text(
        """
        MODEL (
          name sqlmesh_example.incremental_model_with_semicolon,
          kind INCREMENTAL_BY_TIME_RANGE (
            time_column event_date
          ),
          start '2020-01-01',
          cron '@daily',
          grain (id, event_date)
        );

        SELECT
          1 AS id,
          1 AS item_id,
          CAST('2020-01-01' AS DATE) AS event_date
        """
    )

    ctx.load()
    plan = ctx.plan(no_prompts=True, auto_apply=True)

    assert len(plan.context_diff.modified_snapshots) == 1
    assert len(plan.new_snapshots) == 1
    assert plan.new_snapshots[0].is_metadata


def test_invalid_audit_reference():
    sql = """
    MODEL (
      name test,
      audits (not_nulll (columns := (id)))
    );

    SELECT
      1 AS id
    """
    expressions = d.parse(sql)

    with pytest.raises(ConfigError, match="Audit 'not_nulll' is undefined"):
        load_sql_based_model(expressions)


def test_invalid_signal_reference():
    sql = """
    MODEL (
      name test,
      signals (s())
    );

    SELECT
      1 AS id
    """
    expressions = d.parse(sql)

    with pytest.raises(ConfigError, match="Signal 's' is undefined"):
        load_sql_based_model(expressions)


def test_scd_time_data_type_does_not_cause_diff_after_deserialization() -> None:
    for dialect in (
        "athena",
        "bigquery",
        "clickhouse",
        "databricks",
        "duckdb",
        "dune",
        "hive",
        "mysql",
        "postgres",
        "presto",
        "redshift",
        "snowflake",
        "spark",
        "trino",
        "tsql",
    ):
        sql = f"""
        MODEL (
          name test_schema.test_model,
          kind SCD_TYPE_2_BY_COLUMN (
            unique_key ARRAY(col),
            columns ARRAY(col),
            invalidate_hard_deletes TRUE,
            on_destructive_change error
          ),
          dialect {dialect}
        );

        SELECT
          1 AS col
        """

        model = load_sql_based_model(d.parse(sql))
        deserialized_model = SqlModel.parse_raw(model.json())

        assert model.data_hash == deserialized_model.data_hash


def test_python_env_includes_file_path_in_render_definition():
    @model(
        "db.test_model_path",
        kind=dict(
            name=ModelKindName.SCD_TYPE_2_BY_TIME,
            unique_key=["id"],
            disable_restatement=False,
        ),
        columns={"id": "string", "name": "string"},
        optimize_query=False,
    )
    def test_model(context, **kwargs):
        return pd.DataFrame([{"id": context.var("1")}])

    python_model = model.get_registry()["db.test_model_path"].model(
        module_path=Path("."), path=Path("."), dialect="duckdb"
    )

    model_executable_str = python_model.render_definition()[1].sql()
    # Make sure the file path is included in the render definition
    assert "# tests/core/test_model.py" in model_executable_str


def test_resolve_interpolated_variables_when_parsing_python_deps():
    @model(
        name="bla.test_interpolate_var_in_dep_py",
        kind="full",
        columns={'"col"': "int"},
    )
    def unimportant_testing_model(context, **kwargs):
        table1 = context.resolve_table(f"{context.var('schema_name')}.table_name")
        table2 = context.resolve_table(f"{context.blueprint_var('schema_name')}.table_name")

        return context.fetchdf(exp.select("*").from_(table))

    m = model.get_registry()["bla.test_interpolate_var_in_dep_py"].model(
        module_path=Path("."),
        path=Path("."),
        variables={"schema_name": "foo"},
        blueprint_variables={"schema_name": "baz"},
    )

    assert m.depends_on == {'"foo"."table_name"', '"baz"."table_name"'}
    assert m.python_env.get(c.SQLMESH_VARS) == Executable.value({"schema_name": "foo"})
    assert m.python_env.get(c.SQLMESH_BLUEPRINT_VARS) == Executable.value({"schema_name": "baz"})

    @macro()
    def unimportant_testing_macro(evaluator, *projections):
        evaluator.var(f"{evaluator.var('selector')}_variable")
        evaluator.var(f"{evaluator.blueprint_var('selector')}_variable")

        return exp.select(*[f'{p} AS "{p}"' for p in projections])

    m = load_sql_based_model(
        d.parse(
            """
            MODEL (
                name bla.test_interpolate_var_in_dep_sql
            );

            @unimportant_testing_macro();

            SELECT
              1 AS c
            """,
        ),
        variables={"selector": "bla", "bla_variable": 1, "baz_variable": 2},
        blueprint_variables={"selector": "baz"},
    )

    assert m.python_env.get(c.SQLMESH_VARS) == Executable.value(
        {"selector": "bla", "bla_variable": 1, "baz_variable": 2},
        sort_root_dict=True,
    )
    assert m.python_env.get(c.SQLMESH_BLUEPRINT_VARS) == Executable.value(
        {"selector": "baz"}, sort_root_dict=True
    )


def test_extract_schema_in_post_statement(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    config = Config(model_defaults=ModelDefaultsConfig(dialect="duckdb"))

    model1 = tmp_path / "models" / "parent_model.sql"
    model1.parent.mkdir(parents=True, exist_ok=True)
    model1.write_text("MODEL (name x); SELECT 1 AS c")

    model2 = tmp_path / "models" / "child_model.sql"
    model2.parent.mkdir(parents=True, exist_ok=True)
    model2.write_text(
        """
        MODEL (name y);
        SELECT c FROM x;
        ON_VIRTUAL_UPDATE_BEGIN;
        @check_schema('y');
        @check_self_schema();
        ON_VIRTUAL_UPDATE_END;
        """
    )

    check_schema = tmp_path / "macros/check_schema.py"
    check_schema.parent.mkdir(parents=True, exist_ok=True)
    check_schema.write_text("""
from sqlglot import exp
from sqlmesh import macro

@macro()
def check_schema(evaluator, model_name: str):
    if evaluator.runtime_stage != 'loading':
        assert evaluator.columns_to_types(model_name) == {"c": exp.DataType.build("INT")}

@macro()
def check_self_schema(evaluator):
    if evaluator.runtime_stage != 'loading':
        assert evaluator.columns_to_types(evaluator.this_model_fqn) == {"c": exp.DataType.build("INT")}
""")

    context = Context(paths=tmp_path, config=config)
    context.plan(no_prompts=True, auto_apply=True)


def test_model_relies_on_os_getenv(tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    (tmp_path / "macros" / "getenv_macro.py").write_text(
        """
from os import getenv
from sqlmesh import macro

@macro()
def getenv_macro(evaluator):
    getenv("foo", None)
    return 1"""
    )
    (tmp_path / "models" / "model.sql").write_text(
        "MODEL (name test); SELECT @getenv_macro() AS foo"
    )

    monkeypatch.chdir(tmp_path)
    ctx = Context(paths=tmp_path)


def test_invalid_sql_model_query() -> None:
    for kind in ("", ", KIND FULL"):
        expressions = d.parse(
            f"""
            MODEL (name db.table{kind});

            JINJA_STATEMENT_BEGIN;
            SELECT 1 AS c;
            JINJA_END;
            """
        )

        with pytest.raises(
            ConfigError,
            match=r"^A query is required and must be a SELECT statement, a UNION statement, or a JINJA_QUERY block.*",
        ):
            load_sql_based_model(expressions)


def test_query_label_and_authorization_macro() -> None:
    @macro()
    def test_query_label_macro(evaluator):
        return "[('key', 'value')]"

    @macro()
    def test_authorization_macro(evaluator):
        return exp.Literal.string("test_authorization")

    expressions = d.parse(
        """
        MODEL (
           name db.table,
           session_properties (
            query_label = @test_query_label_macro(),
            authorization = @test_authorization_macro()
           )
        );

        SELECT 1 AS c;
        """
    )

    model = load_sql_based_model(expressions)
    assert model.session_properties == {
        "query_label": d.parse_one("@test_query_label_macro()"),
        "authorization": d.parse_one("@test_authorization_macro()"),
    }

    assert model.render_session_properties() == {
        "query_label": d.parse_one("[('key', 'value')]"),
        "authorization": d.parse_one("'test_authorization'"),
    }


def test_boolean_property_validation() -> None:
    expressions = d.parse(
        """
        MODEL (
           name db.table,
           enabled @IF(TRUE, TRUE, FALSE),
           dialect tsql
        );

        SELECT 1 AS c;
        """
    )
    model = load_sql_based_model(expressions, dialect="tsql")
    assert model.enabled


def test_datetime_without_timezone_variable_redshift() -> None:
    expressions = d.parse(
        """
        MODEL (
            name test,
            kind INCREMENTAL_BY_TIME_RANGE (
                time_column test_time_col,
                batch_size 1,
                batch_concurrency 1
            ),
            start '2025-06-01',
            dialect redshift
        );

        SELECT @start_dtntz AS test_time_col
        """
    )
    model = load_sql_based_model(expressions, dialect="redshift")

    assert (
        model.render_query_or_raise().sql("redshift")
        == '''SELECT CAST('1970-01-01 00:00:00' AS TIMESTAMP) AS "test_time_col"'''
    )


def test_python_model_cron_with_blueprints(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    cron_blueprint_model = tmp_path / "models" / "cron_blueprint.py"
    cron_blueprint_model.parent.mkdir(parents=True, exist_ok=True)
    cron_blueprint_model.write_text(
        """
import typing as t
from datetime import datetime

import pandas as pd
from sqlmesh import ExecutionContext, model

@model(
    "@{customer}.some_table",
    kind="FULL",
    cron="@'*/@{min} * * * *'",
    blueprints=[
        {"customer": "customer1", "field_a": "x", "field_b": "y", "min": 5},
        {"customer": "customer2", "field_a": "z", "field_b": "w", "min": 10},
    ],
    columns={
        "field_a": "text",
        "field_b": "text",
        "customer": "text",
    },
    enabled=True
)
def entrypoint(
    context: ExecutionContext,
    start: datetime,
    end: datetime,
    execution_time: datetime,
    **kwargs: t.Any,
) -> pd.DataFrame:
    return pd.DataFrame(
        {
            "field_a": [context.blueprint_var("field_a")],
            "field_b": [context.blueprint_var("field_b")],
            "customer": [context.blueprint_var("customer")],
        }
    )
"""
    )

    context = Context(
        paths=tmp_path, config=Config(model_defaults=ModelDefaultsConfig(dialect="duckdb"))
    )
    models = context.models

    # Test first blueprint
    customer1_model = models.get('"memory"."customer1"."some_table"')
    assert customer1_model is not None
    assert customer1_model.cron == "*/5 * * * *"
    assert customer1_model.enabled
    assert "blueprints" not in customer1_model.all_fields()
    assert customer1_model.python_env.get(c.SQLMESH_BLUEPRINT_VARS) == Executable.value(
        {"customer": "customer1", "field_a": "x", "field_b": "y"}
    )

    # Test second blueprint
    customer2_model = models.get('"memory"."customer2"."some_table"')
    assert customer2_model is not None
    assert customer2_model.cron == "*/10 * * * *"
    assert customer2_model.python_env.get(c.SQLMESH_BLUEPRINT_VARS) == Executable.value(
        {"customer": "customer2", "field_a": "z", "field_b": "w"}
    )

    # Test that the models can be planned and applied
    context.plan(no_prompts=True, auto_apply=True, no_diff=True)

    # Verify the data
    assert context.fetchdf('from "memory"."customer1"."some_table"').to_dict() == {
        "field_a": {0: "x"},
        "field_b": {0: "y"},
        "customer": {0: "customer1"},
    }
    assert context.fetchdf('from "memory"."customer2"."some_table"').to_dict() == {
        "field_a": {0: "z"},
        "field_b": {0: "w"},
        "customer": {0: "customer2"},
    }


def test_python_model_cron_macro_rendering(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    cron_macro_model = tmp_path / "models" / "cron_macro.py"
    cron_macro_model.parent.mkdir(parents=True, exist_ok=True)
    cron_macro_model.write_text(
        """
import pandas as pd
from sqlmesh import model

@model(
    "msc.test_cron_model",
    kind="FULL",
    cron="@{cron_schedule}",
    columns={"a": "int"},
)
def entrypoint(context, **kwargs):
    return pd.DataFrame([{"a": 1}])
"""
    )

    # Test with cron alias
    context_daily = Context(
        paths=tmp_path,
        config=Config(
            model_defaults=ModelDefaultsConfig(dialect="duckdb"),
            variables={"cron_schedule": "@daily"},
        ),
    )
    model_daily = context_daily.models.get('"memory"."msc"."test_cron_model"')

    assert model_daily is not None
    assert model_daily.cron == "@daily"

    # Test with cron expression
    context_expr = Context(
        paths=tmp_path,
        config=Config(
            model_defaults=ModelDefaultsConfig(dialect="duckdb"),
            variables={"cron_schedule": "0 */2 * * *"},
        ),
    )
    model_expr = context_expr.models.get('"memory"."msc"."test_cron_model"')
    assert model_expr is not None
    assert model_expr.cron == "0 */2 * * *"


def test_python_model_normal_cron(tmp_path: Path) -> None:
    init_example_project(tmp_path, engine_type="duckdb", template=ProjectTemplate.EMPTY)

    cron_macro_model = tmp_path / "models" / "cron_macro.py"
    cron_macro_model.parent.mkdir(parents=True, exist_ok=True)
    cron_macro_model.write_text(
        """
import pandas as pd
from sqlmesh import model

@model(
    "msc.normal_test_cron_model",
    kind="FULL",
    cron="@daily",
    columns={"a": "int"},
)
def entrypoint(context, **kwargs):
    return pd.DataFrame([{"a": 1}])
"""
    )

    # Test with cron alias
    context_daily = Context(
        paths=tmp_path,
        config=Config(
            model_defaults=ModelDefaultsConfig(dialect="duckdb"),
            variables={"cron_schedule": "@daily"},
        ),
    )
    model_daily = context_daily.models.get('"memory"."msc"."normal_test_cron_model"')

    assert model_daily is not None
    assert model_daily.cron == "@daily"


def test_render_query_optimize_query_false(assert_exp_eq, sushi_context):
    snapshots = sushi_context.snapshots

    model = sushi_context.get_model("sushi.top_waiters")
    model = model.copy(update={"optimize_query": False})

    upstream_model_version = sushi_context.get_snapshot("sushi.waiter_revenue_by_day").version

    assert_exp_eq(
        model.render_query(snapshots=snapshots).sql(),
        f"""
        WITH "test_macros" AS (
          SELECT
            2 AS "lit_two",
            "revenue" * 2.0 AS "sql_exp",
            CAST("revenue" AS TEXT) AS "sql_lit"
          FROM "memory"."sqlmesh__sushi"."sushi__waiter_revenue_by_day__{upstream_model_version}" AS "waiter_revenue_by_day" /* memory.sushi.waiter_revenue_by_day */
        )
        SELECT
          CAST("waiter_id" AS INT) AS "waiter_id",
          CAST("revenue" AS DOUBLE) AS "revenue"
        FROM "memory"."sqlmesh__sushi"."sushi__waiter_revenue_by_day__{upstream_model_version}" AS "waiter_revenue_by_day" /* memory.sushi.waiter_revenue_by_day */
        WHERE
          "event_date" = (
            SELECT
              MAX("event_date")
            FROM "memory"."sqlmesh__sushi"."sushi__waiter_revenue_by_day__{upstream_model_version}" AS "waiter_revenue_by_day" /* memory.sushi.waiter_revenue_by_day */
          )
        ORDER BY
          "revenue" DESC
        LIMIT 10
        """,
    )


def test_each_macro_with_paren_expression_arg(assert_exp_eq):
    expressions = d.parse(
        """
        MODEL (
            name dataset.@table_name,
            kind VIEW,
            blueprints (
                (
                    table_name := model1,
                    event_columns := (
                        'value' AS property1,
                        'value' AS property2
                    )
                ),
                (
                    table_name := model2,
                    event_columns := (
                        'value' AS property1
                    )
                )
            ),
        );

        SELECT @EACH(@event_columns, x -> x)
        """
    )

    models = load_sql_based_models(expressions, lambda _: {})

    # Should generate 2 models from the blueprints
    assert len(models) == 2

    # Get the models sorted by name for consistent testing
    model1 = next(m for m in models if "model1" in m.name)
    model2 = next(m for m in models if "model2" in m.name)

    # Verify model names
    assert model1.name == "dataset.model1"
    assert model2.name == "dataset.model2"

    assert_exp_eq(
        model1.render_query(),
        """
        SELECT
          'value' AS "property1",
          'value' AS "property2"
        """,
    )

    assert_exp_eq(
        model2.render_query(),
        """
        SELECT
          'value' AS "property1"
        """,
    )


@pytest.mark.parametrize(
    "macro_func, variables",
    [
        ("@M(@v1)", {"v1"}),
        ("@M(@{v1})", {"v1"}),
        ("@M(@SQL('@v1'))", {"v1"}),
        ("@M(@'@{v1}_foo')", {"v1"}),
        ("@M1(@VAR('v1'))", {"v1"}),
        ("@M1(@v1, @M2(@v2), @BLUEPRINT_VAR('v3'))", {"v1", "v2", "v3"}),
        ("@M1(@BLUEPRINT_VAR(@VAR('v1')))", {"v1"}),
    ],
)
def test_extract_macro_func_variable_references(macro_func: str, variables: t.Set[str]) -> None:
    from sqlmesh.core.model.common import _extract_macro_func_variable_references

    macro_func_ast = parse_one(macro_func)
    assert _extract_macro_func_variable_references(macro_func_ast, True)[0] == variables


def test_text_diff_column_descriptions():
    """Test that column_descriptions changes are visible in text_diff."""
    # Create model without column descriptions
    model1 = create_sql_model(
        name="test.model",
        query=parse("SELECT id, name FROM upstream")[0],
    )

    # Create model with column descriptions
    model2 = create_sql_model(
        name="test.model",
        query=parse("SELECT id, name FROM upstream")[0],
        column_descriptions={"id": "User identifier", "name": "User name"},
    )

    # Verify the diff shows the column_descriptions
    diff = model1.text_diff(model2)
    assert diff, "Expected diff to show column_descriptions change"
    assert "+    id = 'User identifier'," in diff
    assert "+    name = 'User name'" in diff

    # Verify reverse diff also works
    diff = model2.text_diff(model1)
    assert diff, "Expected reverse diff to show column_descriptions removal"
    assert "-    id = 'User identifier'," in diff
    assert "-    name = 'User name'" in diff


def test_text_diff_optimize_query():
    """Test that optimize_query changes are visible in text_diff."""
    # Create model without optimize_query
    model1 = create_sql_model(
        name="test.model",
        query=parse("SELECT id, name FROM upstream")[0],
    )

    # Create model with optimize_query enabled
    model2 = create_sql_model(
        name="test.model",
        query=parse("SELECT id, name FROM upstream")[0],
        optimize_query=True,
    )

    # Verify the diff shows the optimize_query change
    diff = model1.text_diff(model2)
    assert diff, "Expected diff to show optimize_query change"
    assert "+  optimize_query" in diff.lower()


def test_raw_jinja_raw_tag():
    expressions = d.parse(
        """
        MODEL (name test);

        JINJA_QUERY_BEGIN;
        SELECT {% raw %} '{{ foo }}' {% endraw %} AS col;
        JINJA_END;
        """
    )

    model = load_sql_based_model(expressions)
    assert model.render_query().sql() == "SELECT '{{ foo }}' AS \"col\""


def test_use_original_sql():
    expressions = d.parse(
        """
        MODEL (name test);

        CREATE TABLE pre (
          a INT
        );

        SELECT
          1,
          2;

        CREATE TABLE post (
          b INT
        );
        """
    )

    model = load_sql_based_model(expressions)
    assert model.query_.sql == "SELECT\n          1,\n          2"
    assert model.pre_statements_[0].sql == "CREATE TABLE pre (\n          a INT\n        )"
    assert model.post_statements_[0].sql == "CREATE TABLE post (\n          b INT\n        );"

    # Now manually create the model and make sure that the original SQL is not used
    model_query = d.parse_one("SELECT 1 AS one")
    assert model_query.meta["sql"] == "SELECT 1 AS one"
    model_query = model_query.select("2 AS two")

    pre_statements = [d.parse_one("CREATE TABLE pre (\n          a INT\n        )")]
    post_statements = [d.parse_one("CREATE TABLE post (\n          b INT\n        );")]

    model = create_sql_model(
        "test",
        model_query,
        pre_statements=pre_statements,
        post_statements=post_statements,
    )
    assert model.query_.sql == "SELECT 1 AS one, 2 AS two"
    assert model.pre_statements_[0].sql == "CREATE TABLE pre (a INT)"
    assert model.post_statements_[0].sql == "CREATE TABLE post (b INT)"