pytrilogy 0.3.148__cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl

trilogy/dialect/base.py

@@ -0,0 +1,1431 @@
+from collections import defaultdict
+from datetime import date, datetime
+from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, Sequence, Union
+
+if TYPE_CHECKING:
+    from trilogy.dialect.config import DialectConfig
+
+from jinja2 import Template
+
+from trilogy.constants import (
+    CONFIG,
+    DEFAULT_NAMESPACE,
+    MagicConstants,
+    Rendering,
+    logger,
+)
+from trilogy.core.constants import UNNEST_NAME
+from trilogy.core.enums import (
+    AddressType,
+    ComparisonOperator,
+    CreateMode,
+    DatePart,
+    FunctionType,
+    GroupMode,
+    Modifier,
+    Ordering,
+    PersistMode,
+    ShowCategory,
+    UnnestMode,
+    WindowType,
+)
+from trilogy.core.internal import DEFAULT_CONCEPTS
+from trilogy.core.models.author import ArgBinding, arg_to_datatype
+from trilogy.core.models.build import (
+    BuildAggregateWrapper,
+    BuildCaseElse,
+    BuildCaseWhen,
+    BuildComparison,
+    BuildConcept,
+    BuildConditional,
+    BuildFilterItem,
+    BuildFunction,
+    BuildMultiSelectLineage,
+    BuildOrderItem,
+    BuildParamaterizedConceptReference,
+    BuildParenthetical,
+    BuildRowsetItem,
+    BuildSubselectComparison,
+    BuildWindowItem,
+)
+from trilogy.core.models.core import (
+    ArrayType,
+    DataType,
+    ListWrapper,
+    MapType,
+    MapWrapper,
+    NumericType,
+    StructType,
+    TraitDataType,
+    TupleWrapper,
+)
+from trilogy.core.models.datasource import Address, Datasource, RawColumnExpr
+from trilogy.core.models.environment import Environment
+from trilogy.core.models.execute import CTE, CompiledCTE, RecursiveCTE, UnionCTE
+from trilogy.core.processing.utility import (
+    decompose_condition,
+    is_scalar_condition,
+    sort_select_output,
+)
+from trilogy.core.query_processor import process_copy, process_persist, process_query
+from trilogy.core.statements.author import (
+    ConceptDeclarationStatement,
+    CopyStatement,
+    CreateStatement,
+    FunctionDeclaration,
+    ImportStatement,
+    MergeStatementV2,
+    MockStatement,
+    MultiSelectStatement,
+    PersistStatement,
+    PublishStatement,
+    RawSQLStatement,
+    RowsetDerivationStatement,
+    SelectStatement,
+    ShowStatement,
+    ValidateStatement,
+)
+from trilogy.core.statements.execute import (
+    PROCESSED_STATEMENT_TYPES,
+    ProcessedCopyStatement,
+    ProcessedCreateStatement,
+    ProcessedMockStatement,
+    ProcessedPublishStatement,
+    ProcessedQuery,
+    ProcessedQueryPersist,
+    ProcessedRawSQLStatement,
+    ProcessedShowStatement,
+    ProcessedStaticValueOutput,
+    ProcessedValidateStatement,
+)
+from trilogy.core.table_processor import (
+    CreateTableInfo,
+    datasource_to_create_table_info,
+    process_create_statement,
+)
+from trilogy.core.utility import safe_quote
+from trilogy.dialect.common import render_join, render_unnest
+from trilogy.hooks.base_hook import BaseHook
+
+
+def null_wrapper(lval: str, rval: str, modifiers: list[Modifier]) -> str:
+
+    if Modifier.NULLABLE in modifiers:
+        return f"({lval} = {rval} or ({lval} is null and {rval} is null))"
+    return f"{lval} = {rval}"
+
+
+LOGGER_PREFIX = "[RENDERING]"
+
+WINDOW_ITEMS = (BuildWindowItem,)
+FILTER_ITEMS = (BuildFilterItem,)
+AGGREGATE_ITEMS = (BuildAggregateWrapper,)
+FUNCTION_ITEMS = (BuildFunction,)
+PARENTHETICAL_ITEMS = (BuildParenthetical,)
+CASE_WHEN_ITEMS = (BuildCaseWhen,)
+CASE_ELSE_ITEMS = (BuildCaseElse,)
+SUBSELECT_COMPARISON_ITEMS = (BuildSubselectComparison,)
+COMPARISON_ITEMS = (BuildComparison,)
+CONDITIONAL_ITEMS = (BuildConditional,)
+
+
+def INVALID_REFERENCE_STRING(x: Any, callsite: str = ""):
+    # if CONFIG.validate_missing:
+    #     raise SyntaxError(f"INVALID_REFERENCE_BUG_{callsite}<{x}>")
+
+    return f"INVALID_REFERENCE_BUG_{callsite}<{x}>"
+
+
+def window_factory(string: str, include_concept: bool = False) -> Callable:
+    def render_window(
+        concept: str, window: str, sort: str, offset: int | None = None
+    ) -> str:
+        if not include_concept:
+            concept = ""
+        if offset is not None:
+            base = f"{string}({concept}, {offset})"
+        else:
+            base = f"{string}({concept})"
+        if window and sort:
+            return f"{base} over (partition by {window} order by {sort} )"
+        elif window:
+            return f"{base} over (partition by {window})"
+        elif sort:
+            return f"{base} over (order by {sort} )"
+        else:
+            return f"{base} over ()"
+
+    return render_window
+
+
+WINDOW_FUNCTION_MAP = {
+    WindowType.LAG: window_factory("lag", include_concept=True),
+    WindowType.LEAD: window_factory("lead", include_concept=True),
+    WindowType.RANK: window_factory("rank"),
+    WindowType.ROW_NUMBER: window_factory("row_number"),
+    WindowType.SUM: window_factory("sum", include_concept=True),
+    WindowType.COUNT: window_factory("count", include_concept=True),
+    WindowType.AVG: window_factory("avg", include_concept=True),
+}
+
+DATATYPE_MAP: dict[DataType, str] = {
+    DataType.STRING: "string",
+    DataType.INTEGER: "int",
+    DataType.FLOAT: "float",
+    DataType.BOOL: "bool",
+    DataType.NUMERIC: "numeric",
+    DataType.MAP: "map",
+    DataType.DATE: "date",
+    DataType.DATETIME: "datetime",
+    DataType.ARRAY: "list",
+}
+
+COMPLEX_DATATYPE_MAP = {
+    DataType.ARRAY: lambda x: f"{x}[]",
+}
+
+
+def render_case(args):
+    return "CASE\n\t" + "\n\t".join(args) + "\n\tEND"
+
+
+def struct_arg(args):
+    return [f"{x[1]}: {x[0]}" for x in zip(args[::2], args[1::2])]
+
+
+def hash_from_args(val, hash_type):
+    hash_type = hash_type[1:-1]
+    if hash_type.lower() == "md5":
+        return f"md5({val})"
+    elif hash_type.lower() == "sha1":
+        return f"sha1({val})"
+    elif hash_type.lower() == "sha256":
+        return f"sha256({val})"
+    elif hash_type.lower() == "sha512":
+        return f"sha512({val})"
+    else:
+        raise ValueError(f"Unsupported hash type: {hash_type}")
+
+
+FUNCTION_MAP = {
+    # generic types
+    FunctionType.ALIAS: lambda x, types: f"{x[0]}",
+    FunctionType.GROUP: lambda x, types: f"{x[0]}",
+    FunctionType.CONSTANT: lambda x, types: f"{x[0]}",
+    FunctionType.TYPED_CONSTANT: lambda x, types: f"{x[0]}",
+    FunctionType.COALESCE: lambda x, types: f"coalesce({','.join(x)})",
+    FunctionType.NULLIF: lambda x, types: f"nullif({x[0]},{x[1]})",
+    FunctionType.CAST: lambda x, types: f"cast({x[0]} as {x[1]})",
+    FunctionType.CASE: lambda x, types: render_case(x),
+    FunctionType.SPLIT: lambda x, types: f"split({x[0]}, {x[1]})",
+    FunctionType.IS_NULL: lambda x, types: f"{x[0]} is null",
+    FunctionType.BOOL: lambda x, types: f"CASE WHEN {x[0]} THEN TRUE ELSE FALSE END",
+    FunctionType.PARENTHETICAL: lambda x, types: f"({x[0]})",
+    # Complex
+    FunctionType.INDEX_ACCESS: lambda x, types: f"{x[0]}[{x[1]}]",
+    FunctionType.MAP_ACCESS: lambda x, types: f"{x[0]}[{x[1]}]",
+    FunctionType.UNNEST: lambda x, types: f"unnest({x[0]})",
+    FunctionType.DATE_SPINE: lambda x, types: f"""unnest(
+        generate_series(
+            {x[0]},
+            {x[1]},
+            INTERVAL '1 day'
+        )
+    )""",
+    FunctionType.RECURSE_EDGE: lambda x, types: f"CASE WHEN {x[1]} IS NULL THEN {x[0]} ELSE {x[1]} END",
+    FunctionType.ATTR_ACCESS: lambda x, types: f"""{x[0]}.{x[1].replace("'", "")}""",
+    FunctionType.STRUCT: lambda x, types: f"{{{', '.join(struct_arg(x))}}}",
+    FunctionType.ARRAY: lambda x, types: f"[{', '.join(x)}]",
+    FunctionType.DATE_LITERAL: lambda x, types: f"date '{x}'",
+    FunctionType.DATETIME_LITERAL: lambda x, types: f"datetime '{x}'",
+    # MAP
+    FunctionType.MAP_KEYS: lambda x, types: f"map_keys({x[0]})",
+    FunctionType.MAP_VALUES: lambda x, types: f"map_values({x[0]})",
+    # ARRAY
+    FunctionType.GENERATE_ARRAY: lambda x, types: f"generate_series({x[0]}, {x[1]}, {x[2]})",
+    FunctionType.ARRAY_SUM: lambda x, types: f"array_sum({x[0]})",
+    FunctionType.ARRAY_DISTINCT: lambda x, types: f"array_distinct({x[0]})",
+    FunctionType.ARRAY_SORT: lambda x, types: f"array_sort({x[0]})",
+    FunctionType.ARRAY_TRANSFORM: lambda args, types: (
+        f"array_transform({args[0]}, {args[1]} -> {args[2]})"
+    ),
+    FunctionType.ARRAY_TO_STRING: lambda args, types: (
+        f"array_to_string({args[0]}, {args[1]})"
+    ),
+    FunctionType.ARRAY_FILTER: lambda args, types: (
+        f"array_filter({args[0]}, {args[1]} -> {args[2]})"
+    ),
+    # math
+    FunctionType.ADD: lambda x, types: " + ".join(x),
+    FunctionType.ABS: lambda x, types: f"abs({x[0]})",
+    FunctionType.SUBTRACT: lambda x, types: " - ".join(x),
+    FunctionType.DIVIDE: lambda x, types: " / ".join(x),
+    FunctionType.MULTIPLY: lambda x, types: " * ".join(x),
+    FunctionType.ROUND: lambda x, types: f"round({x[0]},{x[1]})",
+    FunctionType.FLOOR: lambda x, types: f"floor({x[0]})",
+    FunctionType.CEIL: lambda x, types: f"ceil({x[0]})",
+    FunctionType.MOD: lambda x, types: f"{x[0]} % {x[1]}",
+    FunctionType.POWER: lambda x, types: f"{x[0]} ** {x[1]}",
+    FunctionType.SQRT: lambda x, types: f"sqrt({x[0]})",
+    FunctionType.RANDOM: lambda x, types: "random()",
+    FunctionType.LOG: lambda x, types: (
+        f"log({x[0]})" if x[1] == 10 else f"log({x[0]}, {x[1]})"
+    ),
+    # aggregate types
+    FunctionType.COUNT_DISTINCT: lambda x, types: f"count(distinct {x[0]})",
+    FunctionType.COUNT: lambda x, types: f"count({x[0]})",
+    FunctionType.SUM: lambda x, types: f"sum({x[0]})",
+    FunctionType.ARRAY_AGG: lambda x, types: f"array_agg({x[0]})",
+    FunctionType.LENGTH: lambda x, types: f"length({x[0]})",
+    FunctionType.AVG: lambda x, types: f"avg({x[0]})",
+    FunctionType.MAX: lambda x, types: f"max({x[0]})",
+    FunctionType.MIN: lambda x, types: f"min({x[0]})",
+    FunctionType.ANY: lambda x, types: f"any_value({x[0]})",
+    FunctionType.BOOL_OR: lambda x, types: f"bool_or({x[0]})",
+    FunctionType.BOOL_AND: lambda x, types: f"bool_and({x[0]})",
+    # string types
+    FunctionType.LIKE: lambda x, types: f" {x[0]} like {x[1]} ",
+    FunctionType.UPPER: lambda x, types: f"UPPER({x[0]}) ",
+    FunctionType.LOWER: lambda x, types: f"LOWER({x[0]}) ",
+    FunctionType.SUBSTRING: lambda x, types: f"SUBSTRING({x[0]},{x[1]},{x[2]})",
+    FunctionType.STRPOS: lambda x, types: f"STRPOS({x[0]},{x[1]})",
+    FunctionType.CONTAINS: lambda x, types: f"CONTAINS({x[0]},{x[1]})",
+    FunctionType.REGEXP_CONTAINS: lambda x, types: f"REGEXP_CONTAINS({x[0]},{x[1]})",
+    FunctionType.REGEXP_EXTRACT: lambda x, types: f"REGEXP_EXTRACT({x[0]},{x[1]})",
+    FunctionType.REGEXP_REPLACE: lambda x, types: f"REGEXP_REPLACE({x[0]},{x[1]}, {x[2]})",
+    FunctionType.TRIM: lambda x, types: f"TRIM({x[0]})",
+    FunctionType.REPLACE: lambda x, types: f"REPLACE({x[0]},{x[1]},{x[2]})",
+    FunctionType.HASH: lambda x, types: hash_from_args(x[0], x[1]),
+    # FunctionType.NOT_LIKE: lambda x: f" CASE WHEN {x[0]} like {x[1]} THEN 0 ELSE 1 END",
+    # date types
+    FunctionType.DATE_TRUNCATE: lambda x, types: f"date_trunc({x[0]},{x[1]})",
+    FunctionType.DATE_PART: lambda x, types: f"date_part({x[0]},{x[1]})",
+    FunctionType.DATE_ADD: lambda x, types: f"date_add({x[0]},{x[1]}, {x[2]})",
+    FunctionType.DATE_SUB: lambda x, types: f"date_sub({x[0]},{x[1]}, {x[2]})",
+    FunctionType.DATE_DIFF: lambda x, types: f"date_diff({x[0]},{x[1]}, {x[2]})",
+    FunctionType.DATE: lambda x, types: f"date({x[0]})",
+    FunctionType.DATETIME: lambda x, types: f"datetime({x[0]})",
+    FunctionType.TIMESTAMP: lambda x, types: f"timestamp({x[0]})",
+    FunctionType.SECOND: lambda x, types: f"second({x[0]})",
+    FunctionType.MINUTE: lambda x, types: f"minute({x[0]})",
+    FunctionType.HOUR: lambda x, types: f"hour({x[0]})",
+    FunctionType.DAY: lambda x, types: f"day({x[0]})",
+    FunctionType.DAY_NAME: lambda x, types: f"dayname({x[0]})",
+    FunctionType.DAY_OF_WEEK: lambda x, types: f"day_of_week({x[0]})",
+    FunctionType.WEEK: lambda x, types: f"week({x[0]})",
+    FunctionType.MONTH: lambda x, types: f"month({x[0]})",
+    FunctionType.MONTH_NAME: lambda x, types: f"monthname({x[0]})",
+    FunctionType.QUARTER: lambda x, types: f"quarter({x[0]})",
+    FunctionType.YEAR: lambda x, types: f"year({x[0]})",
+    # string types
+    FunctionType.CONCAT: lambda x, types: f"concat({','.join(x)})",
+    # constant types
+    FunctionType.CURRENT_DATE: lambda x, types: "current_date()",
+    FunctionType.CURRENT_DATETIME: lambda x, types: "current_datetime()",
+}
+
+FUNCTION_GRAIN_MATCH_MAP = {
+    **FUNCTION_MAP,
+    FunctionType.COUNT_DISTINCT: lambda args, types: f"CASE WHEN{args[0]} IS NOT NULL THEN 1 ELSE 0 END",
+    FunctionType.COUNT: lambda args, types: f"CASE WHEN {args[0]} IS NOT NULL THEN 1 ELSE 0 END",
+    FunctionType.SUM: lambda args, types: f"{args[0]}",
+    FunctionType.AVG: lambda args, types: f"{args[0]}",
+    FunctionType.MAX: lambda args, types: f"{args[0]}",
+    FunctionType.MIN: lambda args, types: f"{args[0]}",
+    FunctionType.ANY: lambda args, types: f"{args[0]}",
+}
+
+
+GENERIC_SQL_TEMPLATE: Template = Template(
+    """{%- if ctes %}
+WITH {% if recursive%} RECURSIVE {% endif %}{% for cte in ctes %}
+{{cte.name}} as (
+{{cte.statement}}){% if not loop.last %},{% endif %}{% endfor %}{% endif %}
+{%- if full_select -%}
+{{full_select}}
+{% else -%}
+SELECT
+{%- if limit is not none %}
+TOP {{ limit }}{% endif %}
+{%- for select in select_columns %}
+\t{{ select }}{% if not loop.last %},{% endif %}{% endfor %}
+{% if base %}FROM
+\t{{ base }}{% endif %}{% if joins %}{% for join in joins %}
+\t{{ join }}{% endfor %}{% endif %}{% if where %}
+WHERE
+\t{{ where }}{% endif %}{%- if group_by %}
+GROUP BY {% for group in group_by %}
+\t{{group}}{% if not loop.last %},{% endif %}{% endfor %}{% endif %}{% if having %}
+HAVING
+\t{{ having }}{% endif %}{%- if order_by %}
+ORDER BY{% for order in order_by %}
+\t{{ order }}{% if not loop.last %},{% endif %}{% endfor %}
+{% endif %}{% endif %}
+"""
+)
+
+
+CREATE_TABLE_SQL_TEMPLATE = Template(
+    """
+CREATE {% if create_mode == "create_or_replace" %}OR REPLACE TABLE{% elif create_mode == "create_if_not_exists" %}TABLE IF NOT EXISTS{% else %}TABLE{% endif %} {{ name }} (
+{%- for column in columns %}
+    {{ column.name }} {{ type_map[column.name] }}{% if column.comment %} COMMENT '{{ column.comment }}'{% endif %}{% if not loop.last %},{% endif %}
+{%- endfor %}
+)
+{%- if partition_keys %}
+PARTITIONED BY (
+{%- for partition_key in partition_keys %}
+    {{ partition_key }}{% if not loop.last %},{% endif %}
+{%- endfor %}
+)
+{%- endif %};
+""".strip()
+)
+
+
+def safe_get_cte_value(
+    coalesce: Callable,
+    cte: CTE | UnionCTE,
+    c: BuildConcept,
+    quote_char: str,
+    render_expr: Callable,
+    use_map: dict[str, set[str]],
+) -> Optional[str]:
+    address = c.address
+    raw = cte.source_map.get(address, None)
+
+    if not raw:
+        return None
+    if isinstance(raw, str):
+        rendered = cte.get_alias(c, raw)
+        use_map[raw].add(c.address)
+        return f"{quote_char}{raw}{quote_char}.{safe_quote(rendered, quote_char)}"
+    if isinstance(raw, list) and len(raw) == 1:
+        rendered = cte.get_alias(c, raw[0])
+        if isinstance(rendered, FUNCTION_ITEMS):
+            # if it's a function, we need to render it as a function
+            return f"{render_expr(rendered, cte=cte, raise_invalid=True)}"
+        use_map[raw[0]].add(c.address)
+        return f"{quote_char}{raw[0]}{quote_char}.{safe_quote(rendered, quote_char)}"
+    for x in raw:
+        use_map[x].add(c.address)
+    return coalesce(
+        sorted(
+            [
+                f"{quote_char}{x}{quote_char}.{safe_quote(cte.get_alias(c, x), quote_char)}"
+                for x in raw
+            ]
+        ),
+        [],
+    )
+
+
+class BaseDialect:
+    WINDOW_FUNCTION_MAP = WINDOW_FUNCTION_MAP
+    FUNCTION_MAP = FUNCTION_MAP
+    FUNCTION_GRAIN_MATCH_MAP = FUNCTION_GRAIN_MATCH_MAP
+    QUOTE_CHARACTER = "`"
+    SQL_TEMPLATE = GENERIC_SQL_TEMPLATE
+    CREATE_TABLE_SQL_TEMPLATE = CREATE_TABLE_SQL_TEMPLATE
+    DATATYPE_MAP = DATATYPE_MAP
+    COMPLEX_DATATYPE_MAP = COMPLEX_DATATYPE_MAP
+    UNNEST_MODE = UnnestMode.CROSS_APPLY
+    GROUP_MODE = GroupMode.AUTO
+    EXPLAIN_KEYWORD = "EXPLAIN"
+    NULL_WRAPPER = staticmethod(null_wrapper)
+    ALIAS_ORDER_REFERENCING_ALLOWED = True
+    TABLE_NOT_FOUND_PATTERN: str | None = None  # Dialect-specific pattern to match
+
+    def __init__(
+        self,
+        rendering: Rendering | None = None,
+        config: "DialectConfig | None" = None,
+    ):
+        self.rendering = rendering or CONFIG.rendering
+        self.config = config
+        self.used_map: dict[str, set[str]] = defaultdict(set)
+
+    def render_source(self, address: Address) -> str:
+        if address.type == AddressType.QUERY:
+            return f"({address.location})"
+        if address.is_file:
+            if address.type == AddressType.SQL:
+                with open(address.location, "r", encoding="utf-8") as f:
+                    return f"({f.read()})"
+            raise NotImplementedError(
+                f"File source type {address.type} not supported by this dialect"
+            )
+        return self.safe_quote(address.location)
+
+    def get_table_schema(
+        self, executor, table_name: str, schema: str | None = None
+    ) -> list[tuple]:
+        raise NotImplementedError
+
+    def get_table_primary_keys(
+        self, executor, table_name: str, schema: str | None = None
+    ) -> list[str]:
+        raise NotImplementedError
+
+    def get_table_sample(
+        self,
+        executor,
+        table_name: str,
+        schema: str | None = None,
+        sample_size: int = 10000,
+    ) -> list[tuple]:
+        if schema:
+            qualified_name = f"{schema}.{table_name}"
+        else:
+            qualified_name = table_name
+
+        sample_query = (
+            f"SELECT * FROM {self.safe_quote(qualified_name)} LIMIT {sample_size}"
+        )
+        rows = executor.execute_raw_sql(sample_query).fetchall()
+        return rows
+
+    def get_table_last_modified(
+        self, executor, table_name: str, schema: str | None = None
+    ) -> str | None:
+        from datetime import datetime, timezone
+
+        return datetime.now(timezone.utc).isoformat()
+
+    def hash_column_value(self, column_name: str) -> str:
+        return f"md5(CAST({self.safe_quote(column_name)} AS VARCHAR))"
+
+    def aggregate_checksum(self, hash_expr: str) -> str:
+        return f"BIT_XOR(hash({hash_expr}))"
+
+    def render_order_item(
+        self,
+        order_item: BuildOrderItem,
+        cte: CTE | UnionCTE,
+    ) -> str:
+        if (
+            isinstance(order_item.expr, BuildConcept)
+            and order_item.expr.address in cte.output_columns
+            and self.ALIAS_ORDER_REFERENCING_ALLOWED
+        ):
+            if cte.source_map.get(order_item.expr.address, []):
+                # if it is sourced from somewhere, we need to reference the alias directly
+                return f"{self.render_expr(order_item.expr, cte=cte, )} {order_item.order.value}"
+            # otherwise we've derived it, safe to use alias
+            return f"{self.QUOTE_CHARACTER}{order_item.expr.safe_address}{self.QUOTE_CHARACTER} {order_item.order.value}"
+        return (
+            f"{self.render_expr(order_item.expr, cte=cte, )} {order_item.order.value}"
+        )
+
+    def render_concept_sql(
+        self,
+        c: BuildConcept,
+        cte: CTE | UnionCTE,
+        alias: bool = True,
+        raise_invalid: bool = False,
+    ) -> str:
+        result = None
+        if c.pseudonyms:
+            candidates = [y for y in [cte.get_concept(x) for x in c.pseudonyms] if y]
+            logger.debug(
+                f"{LOGGER_PREFIX} [{c.address}] pseudonym candidates are {[x.address for x in candidates]}"
+            )
+            for candidate in [c] + candidates:
+                try:
+                    logger.debug(
+                        f"{LOGGER_PREFIX} [{c.address}] Attempting rendering w/ candidate {candidate.address}"
+                    )
+                    result = self._render_concept_sql(
+                        candidate,
+                        cte,
+                        raise_invalid=True,
+                    )
+                    if result:
+                        break
+                except ValueError:
+                    continue
+        if not result:
+            result = self._render_concept_sql(
+                c,
+                cte,
+                raise_invalid=raise_invalid,
+            )
+        if alias:
+            return f"{result} as {self.QUOTE_CHARACTER}{c.safe_address}{self.QUOTE_CHARACTER}"
+        return result
+
+    def _render_concept_sql(
+        self,
+        c: BuildConcept,
+        cte: CTE | UnionCTE,
+        raise_invalid: bool = False,
+    ) -> str:
+        # only recurse while it's in sources of the current cte
+        logger.debug(
+            f"{LOGGER_PREFIX} [{c.address}] Starting rendering loop on cte: {cte.name}"
+        )
+
+        # check if it's not inherited AND no pseudonyms are inherited
+        if c.lineage and cte.source_map.get(c.address, []) == []:
+            logger.debug(
+                f"{LOGGER_PREFIX} [{c.address}] rendering concept with lineage that is not already existing"
+            )
+            if isinstance(c.lineage, WINDOW_ITEMS):
+                rendered_order_components = [
+                    f"{self.render_expr(x.expr, cte, raise_invalid=raise_invalid)} {x.order.value}"
+                    for x in c.lineage.order_by
+                ]
+                rendered_over_components = [
+                    self.render_concept_sql(
+                        x, cte, alias=False, raise_invalid=raise_invalid
+                    )
+                    for x in c.lineage.over
+                ]
+
+                rval = self.WINDOW_FUNCTION_MAP[c.lineage.type](
+                    concept=self.render_concept_sql(
+                        c.lineage.content,
+                        cte=cte,
+                        alias=False,
+                        raise_invalid=raise_invalid,
+                    ),
+                    window=",".join(rendered_over_components),
+                    sort=",".join(rendered_order_components),
+                    offset=c.lineage.index,
+                )
+            elif isinstance(c.lineage, FILTER_ITEMS):
+                # for cases when we've optimized this
+                if cte.condition == c.lineage.where.conditional:
+                    rval = self.render_expr(
+                        c.lineage.content, cte=cte, raise_invalid=raise_invalid
+                    )
+                else:
+                    rval = f"CASE WHEN {self.render_expr(c.lineage.where.conditional, cte=cte)} THEN {self.render_expr(c.lineage.content, cte=cte, raise_invalid=raise_invalid)} ELSE NULL END"
+            elif isinstance(c.lineage, BuildRowsetItem):
+                rval = f"{self.render_concept_sql(c.lineage.content, cte=cte, alias=False, raise_invalid=raise_invalid)}"
+            elif isinstance(c.lineage, BuildMultiSelectLineage):
+                if c.address in c.lineage.calculated_derivations:
+                    assert c.lineage.derive is not None
+                    for x in c.lineage.derive.items:
+                        if x.address == c.address:
+                            rval = self.render_expr(
+                                x.expr,
+                                cte=cte,
+                                raise_invalid=raise_invalid,
+                            )
+                            break
+                else:
+                    rval = f"{self.render_concept_sql(c.lineage.find_source(c, cte), cte=cte, alias=False, raise_invalid=raise_invalid)}"
+            elif isinstance(c.lineage, BuildComparison):
+                rval = f"{self.render_expr(c.lineage.left, cte=cte, raise_invalid=raise_invalid)} {c.lineage.operator.value} {self.render_expr(c.lineage.right, cte=cte, raise_invalid=raise_invalid)}"
+            elif isinstance(c.lineage, AGGREGATE_ITEMS):
+                args = [
+                    self.render_expr(v, cte)  # , alias=False)
+                    for v in c.lineage.function.arguments
+                ]
+                if cte.group_to_grain:
+                    rval = self.FUNCTION_MAP[c.lineage.function.operator](args, [])
+                else:
+                    logger.debug(
+                        f"{LOGGER_PREFIX} [{c.address}] ignoring aggregate, already at"
+                        " target grain"
+                    )
+                    rval = f"{self.FUNCTION_GRAIN_MATCH_MAP[c.lineage.function.operator](args, [])}"
+            elif (
+                isinstance(c.lineage, FUNCTION_ITEMS)
+                and c.lineage.operator == FunctionType.UNION
+            ):
+                local_matched = [
+                    x
+                    for x in c.lineage.arguments
+                    if isinstance(x, BuildConcept) and x.address in cte.output_columns
+                ]
+                # if we're sorting by the output of the union
+                if not local_matched:
+                    rval = c.safe_address
+                else:
+                    rval = self.render_expr(local_matched[0], cte)
+            elif (
+                isinstance(c.lineage, FUNCTION_ITEMS)
+                and c.lineage.operator == FunctionType.CONSTANT
+                and self.rendering.parameters is True
+                and c.datatype.data_type != DataType.MAP
+            ):
+                rval = f":{c.safe_address}"
+            else:
+                args = []
+                types = []
+                for arg in c.lineage.arguments:
+                    if (
+                        isinstance(arg, BuildConcept)
+                        and arg.lineage
+                        and isinstance(arg.lineage, FUNCTION_ITEMS)
+                        and arg.lineage.operator
+                        in (
+                            FunctionType.ADD,
+                            FunctionType.SUBTRACT,
+                            FunctionType.DIVIDE,
+                            FunctionType.MULTIPLY,
+                        )
+                    ):
+                        args.append(
+                            self.render_expr(
+                                BuildParenthetical(content=arg),
+                                cte=cte,
+                                raise_invalid=raise_invalid,
+                            )
+                        )
+                    else:
+                        args.append(
+                            self.render_expr(arg, cte=cte, raise_invalid=raise_invalid)
+                        )
+                    types.append(arg_to_datatype(arg))
+
+                if cte.group_to_grain:
+                    rval = f"{self.FUNCTION_MAP[c.lineage.operator](args, types)}"
+                else:
+
+                    rval = f"{self.FUNCTION_GRAIN_MATCH_MAP[c.lineage.operator](args, types)}"
+        else:
+            logger.debug(
+                f"{LOGGER_PREFIX} [{c.address}] Rendering basic lookup from {cte.source_map.get(c.address,None)}"
+            )
+
+            raw_content = cte.get_alias(c)
+            parent = cte.source_map.get(c.address, None)
+            if parent:
+                self.used_map[parent[0]].add(c.address)
+            if isinstance(raw_content, RawColumnExpr):
+                rval = raw_content.text
+            elif isinstance(raw_content, FUNCTION_ITEMS):
+                rval = self.render_expr(
+                    raw_content, cte=cte, raise_invalid=raise_invalid
+                )
+            else:
+                rval = safe_get_cte_value(
+                    self.FUNCTION_MAP[FunctionType.COALESCE],
+                    cte,
+                    c,
+                    self.QUOTE_CHARACTER,
+                    self.render_expr,
+                    self.used_map,
+                )
+                if not rval:
+                    # unions won't have a specific source mapped; just use a generic column reference
+                    # we shouldn't ever have an expression at this point, so will be safe
+                    if isinstance(cte, UnionCTE):
+                        rval = c.safe_address
+                    else:
+                        if raise_invalid:
+                            raise ValueError(
+                                f"Invalid reference string found in query: {rval}, this should never occur. Please report this issue."
+                            )
+                        rval = INVALID_REFERENCE_STRING(
+                            f"Missing source reference to {c.address}"
+                        )
+        return rval
+
+    def render_array_unnest(
+        self,
+        left,
+        right,
+        operator: ComparisonOperator,
+        cte: CTE | UnionCTE | None = None,
+        cte_map: Optional[Dict[str, CTE | UnionCTE]] = None,
+        raise_invalid: bool = False,
+    ):
+        return f"{self.render_expr(left, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} {operator.value} {self.render_expr(right, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)}"
+
+    def render_expr(
+        self,
+        e: Union[
+            BuildConcept,
+            BuildFunction,
+            BuildConditional,
+            BuildAggregateWrapper,
+            BuildComparison,
+            BuildCaseWhen,
+            BuildCaseElse,
+            BuildSubselectComparison,
+            BuildWindowItem,
+            BuildFilterItem,
+            BuildParenthetical,
+            BuildParamaterizedConceptReference,
+            BuildMultiSelectLineage,
+            BuildRowsetItem,
+            str,
+            int,
+            list,
+            bool,
+            float,
+            date,
+            datetime,
+            DataType,
+            TraitDataType,
+            MagicConstants,
+            MapWrapper[Any, Any],
+            MapType,
+            NumericType,
+            StructType,
+            ArrayType,
+            ListWrapper[Any],
+            TupleWrapper[Any],
+            DatePart,
+        ],
+        cte: Optional[CTE | UnionCTE] = None,
+        cte_map: Optional[Dict[str, CTE | UnionCTE]] = None,
+        raise_invalid: bool = False,
+    ) -> str:
+        if isinstance(e, SUBSELECT_COMPARISON_ITEMS):
+            right: Any = e.right
+            while isinstance(right, BuildParenthetical):
+                right = right.content
+            if isinstance(right, BuildConcept):
+                # we won't always have an existnce map
+                # so fall back to the normal map
+                lookup_cte = cte
+                if cte_map and not lookup_cte:
+                    lookup_cte = cte_map.get(right.address)
+
+                assert lookup_cte, "Subselects must be rendered with a CTE in context"
+                if right.address not in lookup_cte.existence_source_map:
+                    lookup = lookup_cte.source_map.get(
+                        right.address,
+                        [
+                            INVALID_REFERENCE_STRING(
+                                f"Missing source reference to {right.address}"
+                            )
+                        ],
+                    )
+                else:
+                    lookup = lookup_cte.existence_source_map[right.address]
+                if len(lookup) > 0:
+                    target = lookup[0]
+                else:
+                    target = INVALID_REFERENCE_STRING(
+                        f"Missing source CTE for {right.address}"
+                    )
+                assert cte, "CTE must be provided for inlined CTEs"
+                self.used_map[target].add(right.address)
+                if target in cte.inlined_ctes:
+                    info = cte.inlined_ctes[target]
+                    return f"{self.render_expr(e.left, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} {e.operator.value} (select {target}.{self.QUOTE_CHARACTER}{right.safe_address}{self.QUOTE_CHARACTER} from {info.new_base} as {target} where {target}.{self.QUOTE_CHARACTER}{right.safe_address}{self.QUOTE_CHARACTER} is not null)"
+                return f"{self.render_expr(e.left, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} {e.operator.value} (select {target}.{self.QUOTE_CHARACTER}{right.safe_address}{self.QUOTE_CHARACTER} from {target} where {target}.{self.QUOTE_CHARACTER}{right.safe_address}{self.QUOTE_CHARACTER} is not null)"
+            elif isinstance(right, BuildParamaterizedConceptReference):
+                if isinstance(right.concept.lineage, BuildFunction) and isinstance(
+                    right.concept.lineage.arguments[0], ListWrapper
+                ):
+                    return self.render_array_unnest(
+                        e.left,
+                        right,
+                        e.operator,
+                        cte=cte,
+                        cte_map=cte_map,
+                        raise_invalid=raise_invalid,
+                    )
+                return f"{self.render_expr(e.left, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} {e.operator.value} {self.render_expr(right, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)}"
+            elif isinstance(
+                right,
+                (ListWrapper, TupleWrapper, BuildParenthetical),
+            ):
+                return f"{self.render_expr(e.left, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} {e.operator.value} {self.render_expr(right, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)}"
+
+            return f"{self.render_expr(e.left, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} {e.operator.value} ({self.render_expr(right, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)})"
+        elif isinstance(e, COMPARISON_ITEMS):
+            return f"{self.render_expr(e.left, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} {e.operator.value} {self.render_expr(e.right, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)}"
+        elif isinstance(e, CONDITIONAL_ITEMS):
+            # conditions need to be nested in parentheses
+            return f"{self.render_expr(e.left, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} {e.operator.value} {self.render_expr(e.right, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)}"
+        elif isinstance(e, WINDOW_ITEMS):
+            rendered_order_components = [
+                f"{self.render_expr(x.expr, cte, cte_map=cte_map, raise_invalid=raise_invalid)} {x.order.value}"
+                for x in e.order_by
+            ]
+            rendered_over_components = [
+                self.render_expr(x, cte, cte_map=cte_map, raise_invalid=raise_invalid)
+                for x in e.over
+            ]
+            return f"{self.WINDOW_FUNCTION_MAP[e.type](concept = self.render_expr(e.content, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid), window=','.join(rendered_over_components), sort=','.join(rendered_order_components))}"  # noqa: E501
+        elif isinstance(e, PARENTHETICAL_ITEMS):
+            # conditions need to be nested in parentheses
+            if isinstance(e.content, list):
+                return f"( {','.join([self.render_expr(x, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid) for x in e.content])} )"
+            return f"( {self.render_expr(e.content, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} )"
+        elif isinstance(e, CASE_WHEN_ITEMS):
+            return f"WHEN {self.render_expr(e.comparison, cte=cte, cte_map=cte_map) } THEN {self.render_expr(e.expr, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid) }"
+        elif isinstance(e, CASE_ELSE_ITEMS):
+            return f"ELSE {self.render_expr(e.expr, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid) }"
+        elif isinstance(e, FUNCTION_ITEMS):
+            arguments = []
+            for arg in e.arguments:
+                if (
+                    isinstance(arg, BuildConcept)
+                    and arg.lineage
+                    and isinstance(arg.lineage, FUNCTION_ITEMS)
+                    and arg.lineage.operator
+                    in (
+                        FunctionType.ADD,
+                        FunctionType.SUBTRACT,
+                        FunctionType.DIVIDE,
+                        FunctionType.MULTIPLY,
+                    )
+                ):
+                    arguments.append(
+                        self.render_expr(
+                            BuildParenthetical(content=arg),
+                            cte=cte,
+                            cte_map=cte_map,
+                            raise_invalid=raise_invalid,
+                        )
+                    )
+                else:
+                    arguments.append(
+                        self.render_expr(
+                            arg, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid
+                        )
+                    )
+
+            if cte and cte.group_to_grain:
+                return self.FUNCTION_MAP[e.operator](arguments, [])
+
+            return self.FUNCTION_GRAIN_MATCH_MAP[e.operator](arguments, [])
+        elif isinstance(e, AGGREGATE_ITEMS):
+            return self.render_expr(
+                e.function, cte, cte_map=cte_map, raise_invalid=raise_invalid
+            )
+        elif isinstance(e, FILTER_ITEMS):
+            return f"CASE WHEN {self.render_expr(e.where.conditional,cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)} THEN {self.render_expr(e.content, cte, cte_map=cte_map, raise_invalid=raise_invalid)} ELSE NULL END"
+        elif isinstance(e, BuildConcept):
+            if (
+                isinstance(e.lineage, FUNCTION_ITEMS)
+                and e.lineage.operator == FunctionType.CONSTANT
+                and self.rendering.parameters is True
+                and e.datatype.data_type != DataType.MAP
+            ):
+                return f":{e.safe_address}"
+            if cte:
+                return self.render_concept_sql(
+                    e,
+                    cte,
+                    alias=False,
+                    raise_invalid=raise_invalid,
+                )
+            elif cte_map:
+                self.used_map[cte_map[e.address].name].add(e.address)
+                return f"{cte_map[e.address].name}.{self.QUOTE_CHARACTER}{e.safe_address}{self.QUOTE_CHARACTER}"
+            return f"{self.QUOTE_CHARACTER}{e.safe_address}{self.QUOTE_CHARACTER}"
+        elif isinstance(e, bool):
+            return f"{e}"
+        elif isinstance(e, str):
+            return f"'{e}'"
+        elif isinstance(e, (int, float)):
+            return str(e)
+        elif isinstance(e, TupleWrapper):
+            return f"({','.join([self.render_expr(x, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid) for x in e])})"
+        elif isinstance(e, MapWrapper):
+            return f"MAP {{{','.join([f'{self.render_expr(k, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)}:{self.render_expr(v, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid)}' for k, v in e.items()])}}}"
+        elif isinstance(e, ListWrapper):
+            return f"{self.FUNCTION_MAP[FunctionType.ARRAY]([self.render_expr(x, cte=cte, cte_map=cte_map, raise_invalid=raise_invalid) for x in e], [])}"
+        elif isinstance(e, DataType):
+            return self.DATATYPE_MAP.get(e, e.value)
+        elif isinstance(e, DatePart):
+            return str(e.value)
+        elif isinstance(e, NumericType):
+            return f"{self.DATATYPE_MAP[DataType.NUMERIC]}({e.precision},{e.scale})"
+        elif isinstance(e, MagicConstants):
+            if e == MagicConstants.NULL:
+                return "null"
+            return str(e.value)
+        elif isinstance(e, date):
+            return self.FUNCTION_MAP[FunctionType.DATE_LITERAL](e, [])
+        elif isinstance(e, datetime):
+            return self.FUNCTION_MAP[FunctionType.DATETIME_LITERAL](e, [])
+        elif isinstance(e, TraitDataType):
+            return self.render_expr(e.type, cte=cte, cte_map=cte_map)
+        elif isinstance(e, ArgBinding):
+            return e.name
+        elif isinstance(e, Ordering):
+            return str(e.value)
+        elif isinstance(e, ArrayType):
+            return f"{self.COMPLEX_DATATYPE_MAP[DataType.ARRAY](self.render_expr(e.value_data_type, cte=cte, cte_map=cte_map))}"
+        elif isinstance(e, list):
+            return f"{self.FUNCTION_MAP[FunctionType.ARRAY]([self.render_expr(x, cte=cte, cte_map=cte_map) for x in e], [])}"
+        elif isinstance(e, BuildParamaterizedConceptReference):
+            if self.rendering.parameters:
+                if e.concept.namespace == DEFAULT_NAMESPACE:
+                    return f":{e.concept.name}"
+                return f":{e.concept.address.replace('.', '_')}"
+            elif e.concept.lineage:
+                return self.render_expr(e.concept.lineage, cte=cte, cte_map=cte_map)
+            return f"{self.QUOTE_CHARACTER}{e.concept.address}{self.QUOTE_CHARACTER}"
+
+        else:
+            raise ValueError(f"Unable to render type {type(e)} {e}")
+
+    def render_cte_group_by(
+        self, cte: CTE | UnionCTE, select_columns
+    ) -> Optional[list[str]]:
+
+        if not cte.group_to_grain:
+            return None
+        base = set(
+            [self.render_concept_sql(c, cte, alias=False) for c in cte.group_concepts]
+        )
+        if self.GROUP_MODE == GroupMode.AUTO:
+            return sorted(list(base))
+
+        else:
+            # find the index of each column in the select columns
+            final = []
+            found = []
+            for idx, c in enumerate(select_columns):
+                pre_alias = c.split(" as ")[0]
+                if pre_alias in base:
+                    final.append(str(idx + 1))
+                    found.append(pre_alias)
+            if not all(c in found for c in base):
+                raise ValueError(
+                    f"Group by columns {base} not found in select columns {select_columns}"
+                )
+            return final
+
+    def safe_quote(self, name: str) -> str:
+        return safe_quote(name, self.QUOTE_CHARACTER)
+
+    def quote(self, name: str) -> str:
+        return f"{self.QUOTE_CHARACTER}{name}{self.QUOTE_CHARACTER}"
+
+    def render_cte(self, cte: CTE | UnionCTE, auto_sort: bool = True) -> CompiledCTE:
+        if isinstance(cte, UnionCTE):
+            base_statement = f"\n{cte.operator}\n".join(
+                [
+                    self.render_cte(child, auto_sort=False).statement
+                    for child in cte.internal_ctes
+                ]
+            )
+            if cte.order_by:
+
+                ordering = [self.render_order_item(i, cte) for i in cte.order_by.items]
+                base_statement += "\nORDER BY " + ",".join(ordering)
+            return CompiledCTE(name=cte.name, statement=base_statement)
+        elif isinstance(cte, RecursiveCTE):
+            base_statement = "\nUNION ALL\n".join(
+                [self.render_cte(child, False).statement for child in cte.internal_ctes]
+            )
+            return CompiledCTE(name=cte.name, statement=base_statement)
+        if self.UNNEST_MODE in (
+            UnnestMode.CROSS_APPLY,
+            UnnestMode.CROSS_JOIN,
+            UnnestMode.CROSS_JOIN_ALIAS,
+            UnnestMode.SNOWFLAKE,
+        ):
+            # for a cross apply, derivation happens in the join
+            # so we only use the alias to select
+            select_columns = [
+                self.render_concept_sql(c, cte)
+                for c in cte.output_columns
+                if c.address not in [y.address for y in cte.join_derived_concepts]
+                and c.address not in cte.hidden_concepts
+            ] + [
+                f"{self.QUOTE_CHARACTER}{c.safe_address}{self.QUOTE_CHARACTER}"
+                for c in cte.join_derived_concepts
+                if c.address not in cte.hidden_concepts
+            ]
+        elif self.UNNEST_MODE in (UnnestMode.CROSS_JOIN_UNNEST, UnnestMode.PRESTO):
+            select_columns = [
+                self.render_concept_sql(c, cte)
+                for c in cte.output_columns
+                if c.address not in [y.address for y in cte.join_derived_concepts]
+                and c.address not in cte.hidden_concepts
+            ] + [
+                f"{UNNEST_NAME} as {self.QUOTE_CHARACTER}{c.safe_address}{self.QUOTE_CHARACTER}"
+                for c in cte.join_derived_concepts
+                if c.address not in cte.hidden_concepts
+            ]
+        else:
+            # otherwse, assume we are unnesting directly in the select
+            select_columns = [
+                self.render_concept_sql(c, cte)
+                for c in cte.output_columns
+                if c.address not in cte.hidden_concepts
+            ]
+        if auto_sort:
+            select_columns = sorted(select_columns, key=lambda x: x)
+        source: str | None = cte.base_name
+        if not cte.render_from_clause:
+            if len(cte.joins) > 0:
+                if cte.join_derived_concepts and self.UNNEST_MODE in (
+                    UnnestMode.CROSS_JOIN_ALIAS,
+                    # UnnestMode.CROSS_JOIN_UNNEST,
+                    UnnestMode.CROSS_JOIN,
+                    UnnestMode.CROSS_APPLY,
+                ):
+
+                    source = f"{render_unnest(self.UNNEST_MODE, self.QUOTE_CHARACTER, cte.join_derived_concepts[0], self.render_expr, cte)}"
+                elif cte.join_derived_concepts and self.UNNEST_MODE in (
+                    UnnestMode.CROSS_JOIN_UNNEST,
+                ):
+                    source = f"{self.render_expr(cte.join_derived_concepts[0], cte)} as {self.QUOTE_CHARACTER}{UNNEST_NAME}{self.QUOTE_CHARACTER}"
+                elif cte.join_derived_concepts and self.UNNEST_MODE in (
+                    UnnestMode.PRESTO,
+                ):
+                    source = f"{self.render_expr(cte.join_derived_concepts[0], cte)} as t({self.QUOTE_CHARACTER}{UNNEST_NAME}{self.QUOTE_CHARACTER})"
+                elif (
+                    cte.join_derived_concepts
+                    and self.UNNEST_MODE == UnnestMode.SNOWFLAKE
+                ):
+                    source = f"{render_unnest(self.UNNEST_MODE, self.QUOTE_CHARACTER, cte.join_derived_concepts[0], self.render_expr, cte)}"
+                # direct - eg DUCK DB - can be directly selected inline
+                elif (
+                    cte.join_derived_concepts and self.UNNEST_MODE == UnnestMode.DIRECT
+                ):
+                    source = None
+                else:
+                    raise SyntaxError("CTE has joins but no from clause")
+            else:
+                source = None
+        else:
+            addr = cte.source_address
+            if isinstance(addr, Address):
+                source = self.render_source(addr)
+            elif cte.quote_address:
+                source = safe_quote(addr, self.QUOTE_CHARACTER)
+            else:
+                source = addr
+            if cte.base_name != cte.base_alias:
+                source = f"{source} as {self.QUOTE_CHARACTER}{cte.base_alias}{self.QUOTE_CHARACTER}"
+        if not cte.render_from_clause:
+            final_joins = []
+        else:
+            final_joins = cte.joins or []
+        where: BuildConditional | BuildParenthetical | BuildComparison | None = None
+        having: BuildConditional | BuildParenthetical | BuildComparison | None = None
+        materialized = {x for x, v in cte.source_map.items() if v}
+        if cte.condition:
+            if not cte.group_to_grain or is_scalar_condition(
+                cte.condition, materialized=materialized
+            ):
+                where = cte.condition
+
+            else:
+                components = decompose_condition(cte.condition)
+                for x in components:
+                    if is_scalar_condition(x, materialized=materialized):
+                        where = where + x if where else x
+                    else:
+                        having = having + x if having else x
+
+        logger.info(f"{LOGGER_PREFIX} {len(final_joins)} joins for cte {cte.name}")
+        return CompiledCTE(
+            name=cte.name,
+            statement=self.SQL_TEMPLATE.render(
+                select_columns=select_columns,
+                base=f"{source}" if source else None,
+                grain=cte.grain,
+                limit=cte.limit,
+                comment=cte.comment if CONFIG.show_comments else None,
+                # some joins may not need to be rendered
+                joins=[
+                    j
+                    for j in [
+                        render_join(
+                            join,
+                            self.QUOTE_CHARACTER,
+                            self.render_expr,
+                            cte,
+                            use_map=self.used_map,
+                            unnest_mode=self.UNNEST_MODE,
+                            null_wrapper=self.NULL_WRAPPER,
+                        )
+                        for join in final_joins
+                    ]
+                    if j
+                ],
+                where=(self.render_expr(where, cte) if where else None),
+                having=(self.render_expr(having, cte) if having else None),
+                order_by=(
+                    [self.render_order_item(i, cte) for i in cte.order_by.items]
+                    if cte.order_by
+                    else None
+                ),
+                group_by=self.render_cte_group_by(cte, select_columns),
+            ),
+        )
+
+    def generate_ctes(
+        self,
+        query: ProcessedQuery,
+    ) -> List[CompiledCTE]:
+        return [self.render_cte(cte) for cte in query.ctes[:-1]] + [
+            # last CTE needs to respect the user output order
+            self.render_cte(sort_select_output(query.ctes[-1], query), auto_sort=False)
+        ]
+
+    def create_show_output(
+        self,
+        environment: Environment,
+        content: ShowCategory,
+    ):
+        if content == ShowCategory.CONCEPTS:
+            output_columns = [
+                environment.concepts[
+                    DEFAULT_CONCEPTS["concept_address"].address
+                ].reference,
+                environment.concepts[
+                    DEFAULT_CONCEPTS["concept_datatype"].address
+                ].reference,
+                environment.concepts[
+                    DEFAULT_CONCEPTS["concept_description"].address
+                ].reference,
+            ]
+            output_values = [
+                {
+                    DEFAULT_CONCEPTS["concept_address"].address: (
+                        concept.name
+                        if concept.namespace == DEFAULT_NAMESPACE
+                        else concept.address
+                    ),
+                    DEFAULT_CONCEPTS["concept_datatype"].address: str(concept.datatype),
+                    DEFAULT_CONCEPTS[
+                        "concept_description"
+                    ].address: concept.metadata.description
+                    or "",
+                }
+                for _, concept in environment.concepts.items()
+                if not concept.is_internal
+            ]
+        else:
+            raise NotImplementedError(f"Show category {content} not implemented")
+        return ProcessedShowStatement(
+            output_columns=output_columns,
+            output_values=[ProcessedStaticValueOutput(values=output_values)],
+        )
+
+    def generate_queries(
+        self,
+        environment: Environment,
+        statements: Sequence[
+            SelectStatement
+            | MultiSelectStatement
+            | PersistStatement
+            | ShowStatement
+            | ConceptDeclarationStatement
+            | RowsetDerivationStatement
+            | ImportStatement
+            | RawSQLStatement
+            | MergeStatementV2
+            | CopyStatement
+            | ValidateStatement
+            | CreateStatement
+            | PublishStatement
+            | MockStatement
+        ],
+        hooks: Optional[List[BaseHook]] = None,
+    ) -> List[PROCESSED_STATEMENT_TYPES]:
+        output: List[PROCESSED_STATEMENT_TYPES] = []
+        for statement in statements:
+            if isinstance(statement, PersistStatement):
+                if hooks:
+                    for hook in hooks:
+                        hook.process_persist_info(statement)
+                persist = process_persist(environment, statement, hooks=hooks)
+                output.append(persist)
+            elif isinstance(statement, CopyStatement):
+                if hooks:
+                    for hook in hooks:
+                        hook.process_select_info(statement.select)
+                copy = process_copy(environment, statement, hooks=hooks)
+                output.append(copy)
+            elif isinstance(statement, SelectStatement):
+                if hooks:
+                    for hook in hooks:
+                        hook.process_select_info(statement)
+                output.append(process_query(environment, statement, hooks=hooks))
+            elif isinstance(statement, MultiSelectStatement):
+                if hooks:
+                    for hook in hooks:
+                        hook.process_multiselect_info(statement)
+                output.append(process_query(environment, statement, hooks=hooks))
+            elif isinstance(statement, RowsetDerivationStatement):
+                if hooks:
+                    for hook in hooks:
+                        hook.process_rowset_info(statement)
+            elif isinstance(statement, ShowStatement):
+                # TODO - encapsulate this a little better
+                if isinstance(statement.content, SelectStatement):
+
+                    output.append(
+                        ProcessedShowStatement(
+                            output_columns=[
+                                environment.concepts[
+                                    DEFAULT_CONCEPTS["query_text"].address
+                                ].reference
+                            ],
+                            output_values=[
+                                process_query(
+                                    environment, statement.content, hooks=hooks
+                                )
+                            ],
+                        )
+                    )
+                elif isinstance(statement.content, ShowCategory):
+                    output.append(
+                        self.create_show_output(environment, statement.content)
+                    )
+                elif isinstance(statement.content, ValidateStatement):
+                    output.append(
+                        ProcessedShowStatement(
+                            output_columns=[
+                                environment.concepts[
+                                    DEFAULT_CONCEPTS["label"].address
+                                ].reference,
+                                environment.concepts[
+                                    DEFAULT_CONCEPTS["query_text"].address
+                                ].reference,
+                                environment.concepts[
+                                    DEFAULT_CONCEPTS["expected"].address
+                                ].reference,
+                            ],
+                            output_values=[
+                                ProcessedValidateStatement(
+                                    scope=statement.content.scope,
+                                    targets=statement.content.targets,
+                                )
+                            ],
+                        )
+                    )
+                else:
+                    raise NotImplementedError(type(statement.content))
+            elif isinstance(statement, RawSQLStatement):
+                output.append(ProcessedRawSQLStatement(text=statement.text))
+            elif isinstance(statement, ValidateStatement):
+                output.append(
+                    ProcessedValidateStatement(
+                        scope=statement.scope,
+                        targets=statement.targets,
+                    )
+                )
+            elif isinstance(statement, MockStatement):
+                output.append(
+                    ProcessedMockStatement(
+                        scope=statement.scope,
+                        targets=statement.targets,
+                    )
+                )
+            elif isinstance(statement, CreateStatement):
+                output.append(process_create_statement(statement, environment))
+            elif isinstance(statement, PublishStatement):
+                output.append(
+                    ProcessedPublishStatement(
+                        scope=statement.scope,
+                        targets=statement.targets,
+                        action=statement.action,
+                    )
+                )
+            elif isinstance(
+                statement,
+                (
+                    ConceptDeclarationStatement,
+                    MergeStatementV2,
+                    ImportStatement,
+                    RowsetDerivationStatement,
+                    Datasource,
+                    FunctionDeclaration,
+                ),
+            ):
+                continue
+            else:
+                raise NotImplementedError(type(statement))
+        return output
+
+    def generate_partitioned_insert(
+        self,
+        query: ProcessedQueryPersist,
+        recursive: bool,
+        compiled_ctes: list[CompiledCTE],
+    ) -> str:
+        return self.SQL_TEMPLATE.render(
+            recursive=recursive,
+            output=f"INSERT OVERWRITE {self.safe_quote(query.output_to.address.location)}",
+            full_select=compiled_ctes[-1].statement,
+            ctes=compiled_ctes[:-1],
+        )
+
+    def compile_create_table_statement(
+        self, target: CreateTableInfo, create_mode: CreateMode
+    ) -> str:
+        type_map = {}
+        for c in target.columns:
+            type_map[c.name] = self.render_expr(c.type)
+        return self.CREATE_TABLE_SQL_TEMPLATE.render(
+            create_mode=create_mode.value,
+            name=self.safe_quote(target.name),
+            columns=target.columns,
+            type_map=type_map,
+            partition_keys=target.partition_keys,
+        )
+
+    def compile_statement(
+        self,
+        query: PROCESSED_STATEMENT_TYPES,
+    ) -> str:
+        if isinstance(query, ProcessedShowStatement):
+            return ";\n".join(
+                [
+                    f"{self.EXPLAIN_KEYWORD} {self.compile_statement(x)}"
+                    for x in query.output_values
+                    if isinstance(x, (ProcessedQuery, ProcessedCopyStatement))
+                ]
+            )
+        elif isinstance(query, ProcessedRawSQLStatement):
+            return query.text
+
+        elif isinstance(query, ProcessedValidateStatement):
+            return "--Trilogy validate statements do not have a generic SQL representation;\nselect 1;"
+        elif isinstance(query, ProcessedMockStatement):
+            return "--Trilogy mock statements do not have a generic SQL representation;\nselect 1;"
+        elif isinstance(query, ProcessedPublishStatement):
+            return "--Trilogy publish statements do not have a generic SQL representation;\nselect 1;"
+        elif isinstance(query, ProcessedCreateStatement):
+
+            text = []
+            for target in query.targets:
+                text.append(
+                    self.compile_create_table_statement(target, query.create_mode)
+                )
+            return "\n".join(text)
+
+        recursive = any(isinstance(x, RecursiveCTE) for x in query.ctes)
+
+        compiled_ctes = self.generate_ctes(query)
+        output = None
+        if isinstance(query, ProcessedQueryPersist):
+            if query.persist_mode == PersistMode.OVERWRITE:
+                create_table_info = datasource_to_create_table_info(query.datasource)
+                output = f"{self.compile_create_table_statement(create_table_info, CreateMode.CREATE_OR_REPLACE)} INSERT INTO {self.safe_quote(query.output_to.address.location)} "
+            elif query.persist_mode == PersistMode.APPEND:
+                if query.partition_by:
+                    return self.generate_partitioned_insert(
+                        query, recursive, compiled_ctes
+                    )
+                else:
+                    output = f"INSERT INTO {self.safe_quote(query.output_to.address.location)} "
+            else:
+                raise NotImplementedError(
+                    f"Persist mode {query.persist_mode} not implemented"
+                )
+
+        final = self.SQL_TEMPLATE.render(
+            recursive=recursive,
+            output=output,
+            full_select=compiled_ctes[-1].statement,
+            ctes=compiled_ctes[:-1],
+        )
+
+        if CONFIG.strict_mode and INVALID_REFERENCE_STRING(1) in final:
+            raise ValueError(
+                f"Invalid reference string found in query: {final}, this should never"
+                " occur. Please create an issue to report this."
+            )
+        logger.info(f"{LOGGER_PREFIX} Compiled query: {final}")
+        return final