pytrilogy 0.3.142__cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl

trilogy/core/functions.py

@@ -0,0 +1,1227 @@
+from dataclasses import dataclass
+from datetime import date, datetime
+from typing import Any, Callable, Optional
+
+from lark.tree import Meta
+
+from trilogy.constants import MagicConstants
+from trilogy.core.enums import (
+    DatePart,
+    FunctionClass,
+    FunctionType,
+    Granularity,
+    InfiniteFunctionArgs,
+    Purpose,
+)
+from trilogy.core.exceptions import InvalidSyntaxException
+from trilogy.core.models.author import (
+    AggregateWrapper,
+    Concept,
+    ConceptRef,
+    Conditional,
+    CustomType,
+    Function,
+    Parenthetical,
+    UndefinedConcept,
+    WindowItem,
+)
+from trilogy.core.models.core import (
+    CONCRETE_TYPES,
+    ArrayType,
+    DataType,
+    MapType,
+    NumericType,
+    StructType,
+    TraitDataType,
+    arg_to_datatype,
+    merge_datatypes,
+)
+from trilogy.core.models.environment import Environment
+
+GENERIC_ARGS = Concept | ConceptRef | Function | str | int | float | date | datetime
+
+
+CUSTOM_PLACEHOLDER = CustomType(
+    name="__placeholder__", type=DataType.UNKNOWN, drop_on=[], add_on=[]
+)
+
+
+@dataclass
+class FunctionConfig:
+    arg_count: int = 1
+    valid_inputs: set[DataType] | list[set[DataType]] | None = None
+    output_purpose: Purpose | None = None
+    output_type: (
+        DataType | ArrayType | MapType | StructType | NumericType | TraitDataType | None
+    ) = None
+    output_type_function: Optional[Callable] = None
+
+
+def get_unnest_output_type(args: list[Any]) -> CONCRETE_TYPES:
+    output = arg_to_datatype(args[0])
+    if isinstance(output, (ArrayType, MapType)):
+        output = output.value_data_type
+    else:
+        output = DataType.STRING
+    return output
+
+
+def get_coalesce_output_type(args: list[Any]) -> CONCRETE_TYPES:
+    non_null = [x for x in args if not x == MagicConstants.NULL]
+    processed = [arg_to_datatype(x) for x in non_null if x]
+    if not len(set(processed)) == 1:
+        raise InvalidSyntaxException(
+            f"All arguments to coalesce must be of the same type, have {set(arg_to_datatype(x) for x in args)}"
+        )
+    return processed[0]
+
+
+def get_transform_output_type(args: list[Any]) -> CONCRETE_TYPES:
+    return ArrayType(type=arg_to_datatype(args[2]))
+
+
+def get_index_output_type(
+    args: list[Any],
+) -> CONCRETE_TYPES:
+    arg = args[0]
+    datatype = arg_to_datatype(arg)
+    if isinstance(datatype, ArrayType):
+        return datatype.value_data_type
+    elif isinstance(datatype, MapType):
+        return datatype.value_data_type
+    return datatype
+
+
+def get_attr_datatype(
+    args: list[Any],
+) -> CONCRETE_TYPES:
+    arg = args[0]
+    lookup = args[1]
+    datatype = arg_to_datatype(arg)
+    if isinstance(datatype, StructType):
+        return datatype.field_types[lookup]
+    return datatype
+
+
+def get_cast_output_type(
+    args: list[Any],
+) -> DataType | TraitDataType:
+    base = arg_to_datatype(args[0])
+    if isinstance(base, TraitDataType):
+        traits = base.traits
+    else:
+        traits = []
+    if isinstance(args[1], TraitDataType):
+        return TraitDataType(
+            type=args[1].type, traits=list(set(traits + args[1].traits))
+        )
+    elif traits:
+        return TraitDataType(type=args[1], traits=traits)
+    return args[1]
+
+
+def get_output_type_at_index(args, index: int):
+    return arg_to_datatype(args[index])
+
+
+def validate_case_output(
+    args: list[Any],
+) -> DataType:
+    datatypes = set()
+    mapz = dict()
+    for arg in args:
+        output_datatype = arg_to_datatype(arg.expr)
+        if output_datatype != DataType.NULL:
+            datatypes.add(output_datatype.data_type)
+        mapz[str(arg.expr)] = output_datatype
+    known = [x for x in datatypes if x != DataType.UNKNOWN]
+    if len(known) == 0:
+        return DataType.UNKNOWN
+    if not len(known) == 1:
+        raise SyntaxError(
+            f"All case expressions must have the same output datatype, got {datatypes} from {mapz}"
+        )
+    return known.pop()
+
+
+def create_struct_output(
+    args: list[Any],
+) -> StructType:
+    zipped = dict(zip(args[1::2], args[::2]))
+    types = [arg_to_datatype(x) for x in args[::2]]
+    return StructType(fields=types, fields_map=zipped)
+
+
+def get_date_part_output(args: list[Any]):
+    target = args[1]
+    if target == DatePart.YEAR:
+        return TraitDataType(type=DataType.INTEGER, traits=["year"])
+    elif target == DatePart.MONTH:
+        return TraitDataType(type=DataType.INTEGER, traits=["month"])
+    elif target == DatePart.DAY:
+        return TraitDataType(type=DataType.INTEGER, traits=["day"])
+    elif target == DatePart.HOUR:
+        return TraitDataType(type=DataType.INTEGER, traits=["hour"])
+    elif target == DatePart.MINUTE:
+        return TraitDataType(type=DataType.INTEGER, traits=["minute"])
+    elif target == DatePart.SECOND:
+        return TraitDataType(type=DataType.INTEGER, traits=["second"])
+    elif target == DatePart.WEEK:
+        return TraitDataType(type=DataType.INTEGER, traits=["week"])
+    elif target == DatePart.QUARTER:
+        return TraitDataType(type=DataType.INTEGER, traits=["quarter"])
+    elif target == DatePart.DAY_OF_WEEK:
+        return TraitDataType(type=DataType.INTEGER, traits=["day_of_week"])
+    else:
+        raise InvalidSyntaxException(f"Date part not supported for {target}")
+
+
+def get_date_trunc_output(
+    args: list[Any],
+):
+    target: DatePart = args[1]
+    if target == DatePart.YEAR:
+        return DataType.DATE
+    elif target == DatePart.MONTH:
+        return DataType.DATE
+    elif target == DatePart.DAY:
+        return DataType.DATE
+    elif target == DatePart.HOUR:
+        return DataType.DATETIME
+    elif target == DatePart.MINUTE:
+        return DataType.DATETIME
+    elif target == DatePart.SECOND:
+        return DataType.DATETIME
+    elif target == DatePart.WEEK:
+        return DataType.DATE
+    elif target == DatePart.QUARTER:
+        return DataType.DATE
+    else:
+        raise InvalidSyntaxException(f"Date truncation not supported for {target}")
+
+
+def get_map_key_type(arg):
+    arg_datatype = arg_to_datatype(arg)
+    if isinstance(arg_datatype, MapType):
+        return ArrayType(type=arg_datatype.key_data_type)
+    return ArrayType(type=DataType.STRING)
+
+
+def get_map_value_type(arg):
+    arg_datatype = arg_to_datatype(arg)
+    if isinstance(arg_datatype, MapType):
+        return ArrayType(type=arg_datatype.value_data_type)
+    return ArrayType(type=DataType.STRING)
+
+
+FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
+    FunctionType.ALIAS: FunctionConfig(
+        arg_count=1,
+    ),
+    FunctionType.PARENTHETICAL: FunctionConfig(
+        arg_count=1,
+    ),
+    FunctionType.UNNEST: FunctionConfig(
+        valid_inputs={
+            DataType.ARRAY,
+        },
+        output_purpose=Purpose.KEY,
+        output_type_function=get_unnest_output_type,
+        arg_count=1,
+    ),
+    FunctionType.DATE_SPINE: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+        },
+        output_purpose=Purpose.KEY,
+        output_type=DataType.DATE,
+        arg_count=2,
+    ),
+    FunctionType.RECURSE_EDGE: FunctionConfig(
+        arg_count=2,
+    ),
+    FunctionType.GROUP: FunctionConfig(
+        arg_count=-1,
+        output_type_function=lambda args: get_output_type_at_index(args, 0),
+    ),
+    FunctionType.COUNT: FunctionConfig(
+        output_purpose=Purpose.METRIC,
+        output_type=DataType.INTEGER,
+        arg_count=1,
+    ),
+    FunctionType.COUNT_DISTINCT: FunctionConfig(
+        output_purpose=Purpose.METRIC,
+        output_type=DataType.INTEGER,
+        arg_count=1,
+    ),
+    FunctionType.MAX: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.FLOAT,
+            DataType.NUMBER,
+            DataType.DATE,
+            DataType.DATETIME,
+            DataType.TIMESTAMP,
+            DataType.BOOL,
+        },
+        output_purpose=Purpose.METRIC,
+        arg_count=1,
+    ),
+    FunctionType.MIN: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.FLOAT,
+            DataType.NUMBER,
+            DataType.DATE,
+            DataType.DATETIME,
+            DataType.TIMESTAMP,
+        },
+        output_purpose=Purpose.METRIC,
+        arg_count=1,
+    ),
+    FunctionType.SPLIT: FunctionConfig(
+        valid_inputs={DataType.STRING},
+        output_purpose=Purpose.PROPERTY,
+        output_type=ArrayType(type=DataType.STRING),
+        arg_count=2,
+    ),
+    FunctionType.INDEX_ACCESS: FunctionConfig(
+        valid_inputs=[
+            {
+                DataType.ARRAY,
+            },
+            {
+                DataType.INTEGER,
+            },
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=get_index_output_type,
+        arg_count=2,
+    ),
+    FunctionType.MAP_KEYS: FunctionConfig(
+        valid_inputs={
+            DataType.MAP,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=lambda args: get_map_key_type(args[0]),
+        arg_count=1,
+    ),
+    FunctionType.MAP_VALUES: FunctionConfig(
+        valid_inputs={
+            DataType.MAP,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=lambda args: get_map_value_type(args[0]),
+        arg_count=1,
+    ),
+    FunctionType.GENERATE_ARRAY: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.INTEGER,
+            DataType.INTEGER,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=ArrayType(type=DataType.INTEGER),
+        arg_count=3,
+    ),
+    FunctionType.ARRAY_DISTINCT: FunctionConfig(
+        valid_inputs={
+            DataType.ARRAY,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=lambda args: get_output_type_at_index(args, 0),
+        arg_count=1,
+    ),
+    FunctionType.ARRAY_SORT: FunctionConfig(
+        valid_inputs=[
+            {DataType.ARRAY},
+            {DataType.STRING},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=lambda args: get_output_type_at_index(args, 0),
+        arg_count=2,
+    ),
+    FunctionType.ARRAY_TRANSFORM: FunctionConfig(
+        valid_inputs=[
+            {
+                DataType.ARRAY,
+            },
+            {*DataType},
+            {*DataType},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=get_transform_output_type,
+        arg_count=3,
+    ),
+    FunctionType.ARRAY_FILTER: FunctionConfig(
+        valid_inputs=[
+            {
+                DataType.ARRAY,
+            },
+            {*DataType},
+            {*DataType},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=get_transform_output_type,
+        arg_count=3,
+    ),
+    FunctionType.ARRAY_TO_STRING: FunctionConfig(
+        valid_inputs={
+            DataType.ARRAY,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=2,
+    ),
+    FunctionType.ARRAY_SUM: FunctionConfig(
+        valid_inputs={
+            DataType.ARRAY,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=get_index_output_type,
+        arg_count=1,
+    ),
+    FunctionType.MAP_ACCESS: FunctionConfig(
+        valid_inputs=[
+            {
+                DataType.MAP,
+            },
+            {
+                DataType.INTEGER,
+                DataType.STRING,
+            },
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=get_index_output_type,
+        arg_count=2,
+    ),
+    FunctionType.ATTR_ACCESS: FunctionConfig(
+        valid_inputs=[
+            {DataType.STRUCT},
+            {
+                DataType.STRING,
+            },
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=get_attr_datatype,
+        arg_count=2,
+    ),
+    FunctionType.ABS: FunctionConfig(
+        valid_inputs={DataType.INTEGER, DataType.FLOAT, DataType.NUMBER},
+        output_purpose=Purpose.PROPERTY,
+        arg_count=1,
+    ),
+    FunctionType.NULLIF: FunctionConfig(
+        valid_inputs={*DataType},
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=lambda args: get_output_type_at_index(args, 0),
+        arg_count=2,
+    ),
+    FunctionType.COALESCE: FunctionConfig(
+        valid_inputs={*DataType},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.INTEGER,
+        arg_count=-1,
+        output_type_function=get_coalesce_output_type,
+    ),
+    FunctionType.CURRENT_DATE: FunctionConfig(
+        output_purpose=Purpose.CONSTANT,
+        output_type=DataType.DATE,
+        arg_count=0,
+    ),
+    FunctionType.CURRENT_DATETIME: FunctionConfig(
+        output_purpose=Purpose.CONSTANT,
+        output_type=DataType.DATETIME,
+        arg_count=0,
+    ),
+    FunctionType.CURRENT_TIMESTAMP: FunctionConfig(
+        output_purpose=Purpose.CONSTANT,
+        output_type=DataType.TIMESTAMP,
+        arg_count=0,
+    ),
+    FunctionType.BOOL: FunctionConfig(
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.BOOL,
+        arg_count=1,
+    ),
+    FunctionType.STRPOS: FunctionConfig(
+        valid_inputs=[
+            {DataType.STRING},
+            {DataType.STRING},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.INTEGER,
+        arg_count=2,
+    ),
+    FunctionType.REPLACE: FunctionConfig(
+        valid_inputs=[
+            {DataType.STRING},
+            {DataType.STRING},
+            {DataType.STRING},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=3,
+    ),
+    FunctionType.CONTAINS: FunctionConfig(
+        valid_inputs=[
+            {DataType.STRING},
+            {DataType.STRING},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.BOOL,
+        arg_count=2,
+    ),
+    FunctionType.TRIM: FunctionConfig(
+        valid_inputs=[
+            {DataType.STRING},
+            {DataType.STRING},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=2,
+    ),
+    FunctionType.SUBSTRING: FunctionConfig(
+        valid_inputs=[{DataType.STRING}, {DataType.INTEGER}, {DataType.INTEGER}],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=3,
+    ),
+    FunctionType.UNION: FunctionConfig(
+        valid_inputs={*DataType},
+        output_purpose=Purpose.KEY,
+        arg_count=-1,
+    ),
+    FunctionType.LIKE: FunctionConfig(
+        valid_inputs={DataType.STRING},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.BOOL,
+        arg_count=2,
+    ),
+    FunctionType.ILIKE: FunctionConfig(
+        valid_inputs={DataType.STRING},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.BOOL,
+        arg_count=2,
+    ),
+    FunctionType.UPPER: FunctionConfig(
+        valid_inputs={DataType.STRING},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=1,
+    ),
+    FunctionType.LOWER: FunctionConfig(
+        valid_inputs={DataType.STRING},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=1,
+    ),
+    FunctionType.REGEXP_CONTAINS: FunctionConfig(
+        valid_inputs={DataType.STRING},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.BOOL,
+        arg_count=2,
+    ),
+    FunctionType.REGEXP_EXTRACT: FunctionConfig(
+        valid_inputs=[{DataType.STRING}, {DataType.STRING}, {DataType.INTEGER}],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=3,
+    ),
+    FunctionType.REGEXP_REPLACE: FunctionConfig(
+        valid_inputs={DataType.STRING},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=3,
+    ),
+    FunctionType.DATE: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.DATE,
+        arg_count=1,
+    ),
+    FunctionType.DATE_TRUNCATE: FunctionConfig(
+        valid_inputs=[
+            {
+                DataType.DATE,
+                DataType.TIMESTAMP,
+                DataType.DATETIME,
+                DataType.STRING,
+            },
+            {DataType.DATE_PART},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        # output_type=DataType.DATE,
+        output_type_function=get_date_trunc_output,
+        arg_count=2,
+    ),
+    FunctionType.DATE_PART: FunctionConfig(
+        valid_inputs=[
+            {
+                DataType.DATE,
+                DataType.TIMESTAMP,
+                DataType.DATETIME,
+                DataType.STRING,
+            },
+            {DataType.DATE_PART},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=get_date_part_output,
+        arg_count=2,
+    ),
+    FunctionType.DATE_ADD: FunctionConfig(
+        valid_inputs=[
+            {
+                DataType.DATE,
+                DataType.TIMESTAMP,
+                DataType.DATETIME,
+                DataType.STRING,
+            },
+            {DataType.DATE_PART},
+            {DataType.INTEGER},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.DATE,
+        arg_count=3,
+    ),
+    FunctionType.DATE_SUB: FunctionConfig(
+        valid_inputs=[
+            {
+                DataType.DATE,
+                DataType.TIMESTAMP,
+                DataType.DATETIME,
+                DataType.STRING,
+            },
+            {DataType.DATE_PART},
+            {DataType.INTEGER},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.DATE,
+        arg_count=3,
+    ),
+    FunctionType.DATE_DIFF: FunctionConfig(
+        valid_inputs=[
+            {
+                DataType.DATE,
+                DataType.TIMESTAMP,
+                DataType.DATETIME,
+                DataType.STRING,
+            },
+            {
+                DataType.DATE,
+                DataType.TIMESTAMP,
+                DataType.DATETIME,
+                DataType.STRING,
+            },
+            {DataType.DATE_PART},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.INTEGER,
+        arg_count=3,
+    ),
+    FunctionType.DATETIME: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.DATETIME,
+        arg_count=1,
+    ),
+    FunctionType.TIMESTAMP: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.TIMESTAMP,
+        arg_count=1,
+    ),
+    FunctionType.SECOND: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.INTEGER, traits=["second"]),
+        arg_count=1,
+    ),
+    FunctionType.MINUTE: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.INTEGER, traits=["minute"]),
+        arg_count=1,
+    ),
+    FunctionType.HOUR: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.INTEGER, traits=["hour"]),
+        arg_count=1,
+    ),
+    FunctionType.DAY: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.INTEGER, traits=["day"]),
+        arg_count=1,
+    ),
+    FunctionType.DAY_NAME: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            # DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.STRING, traits=["day_name"]),
+        arg_count=1,
+    ),
+    FunctionType.WEEK: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.INTEGER, traits=["week"]),
+        arg_count=1,
+    ),
+    FunctionType.MONTH: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.INTEGER, traits=["month"]),
+        arg_count=1,
+    ),
+    FunctionType.MONTH_NAME: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            # DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.STRING, traits=["month_name"]),
+        arg_count=1,
+    ),
+    FunctionType.QUARTER: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.INTEGER, traits=["quarter"]),
+        arg_count=1,
+    ),
+    FunctionType.YEAR: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.INTEGER, traits=["year"]),
+        arg_count=1,
+    ),
+    FunctionType.DAY_OF_WEEK: FunctionConfig(
+        valid_inputs={
+            DataType.DATE,
+            DataType.TIMESTAMP,
+            DataType.DATETIME,
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=TraitDataType(type=DataType.INTEGER, traits=["day_of_week"]),
+        arg_count=1,
+    ),
+    FunctionType.ADD: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.FLOAT,
+            DataType.NUMBER,
+            DataType.NUMERIC,
+        },
+        output_purpose=Purpose.PROPERTY,
+        arg_count=InfiniteFunctionArgs,
+    ),
+    FunctionType.SUBTRACT: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.FLOAT,
+            DataType.NUMBER,
+            DataType.NUMERIC,
+        },
+        output_purpose=Purpose.PROPERTY,
+        arg_count=InfiniteFunctionArgs,
+    ),
+    FunctionType.MULTIPLY: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.FLOAT,
+            DataType.NUMBER,
+            DataType.NUMERIC,
+        },
+        output_purpose=Purpose.PROPERTY,
+        arg_count=InfiniteFunctionArgs,
+    ),
+    FunctionType.POWER: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.FLOAT,
+            DataType.NUMBER,
+            DataType.NUMERIC,
+        },
+        output_purpose=Purpose.PROPERTY,
+        arg_count=2,
+    ),
+    FunctionType.DIVIDE: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.FLOAT,
+            DataType.NUMBER,
+            DataType.NUMERIC,
+        },
+        output_purpose=Purpose.PROPERTY,
+        arg_count=InfiniteFunctionArgs,
+    ),
+    FunctionType.MOD: FunctionConfig(
+        valid_inputs=[
+            {DataType.INTEGER, DataType.FLOAT, DataType.NUMBER, DataType.NUMERIC},
+            {DataType.INTEGER},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.INTEGER,
+        arg_count=2,
+    ),
+    FunctionType.SQRT: FunctionConfig(
+        valid_inputs=[
+            {DataType.INTEGER, DataType.FLOAT, DataType.NUMBER, DataType.NUMERIC},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.INTEGER,
+        arg_count=1,
+    ),
+    FunctionType.LOG: FunctionConfig(
+        valid_inputs=[
+            {DataType.INTEGER, DataType.FLOAT, DataType.NUMBER, DataType.NUMERIC},
+            {DataType.INTEGER},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.FLOAT,
+        arg_count=2,
+    ),
+    FunctionType.RANDOM: FunctionConfig(
+        valid_inputs=[],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.FLOAT,
+        arg_count=1,
+    ),
+    FunctionType.ROUND: FunctionConfig(
+        valid_inputs=[
+            {DataType.INTEGER, DataType.FLOAT, DataType.NUMBER, DataType.NUMERIC},
+            {DataType.INTEGER},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=lambda args: get_output_type_at_index(args, 0),
+        arg_count=2,
+    ),
+    FunctionType.FLOOR: FunctionConfig(
+        valid_inputs=[
+            {DataType.INTEGER, DataType.FLOAT, DataType.NUMBER, DataType.NUMERIC},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.INTEGER,
+        arg_count=1,
+    ),
+    FunctionType.CEIL: FunctionConfig(
+        valid_inputs=[
+            {DataType.INTEGER, DataType.FLOAT, DataType.NUMBER, DataType.NUMERIC},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.INTEGER,
+        arg_count=1,
+    ),
+    FunctionType.CUSTOM: FunctionConfig(
+        output_purpose=Purpose.PROPERTY,
+        arg_count=InfiniteFunctionArgs,
+    ),
+    FunctionType.CASE: FunctionConfig(
+        output_purpose=Purpose.PROPERTY,
+        output_type_function=validate_case_output,
+        arg_count=InfiniteFunctionArgs,
+    ),
+    FunctionType.CAST: FunctionConfig(
+        output_purpose=Purpose.PROPERTY,
+        arg_count=2,
+        output_type_function=get_cast_output_type,
+    ),
+    FunctionType.CONCAT: FunctionConfig(
+        valid_inputs={DataType.STRING},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=InfiniteFunctionArgs,
+    ),
+    FunctionType.CONSTANT: FunctionConfig(
+        output_purpose=Purpose.CONSTANT,
+        arg_count=1,
+    ),
+    FunctionType.TYPED_CONSTANT: FunctionConfig(
+        output_purpose=Purpose.CONSTANT,
+        output_type_function=get_cast_output_type,
+        arg_count=2,
+    ),
+    FunctionType.IS_NULL: FunctionConfig(
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.BOOL,
+        arg_count=1,
+    ),
+    FunctionType.STRUCT: FunctionConfig(
+        output_purpose=Purpose.PROPERTY,
+        arg_count=InfiniteFunctionArgs,
+        output_type_function=create_struct_output,
+    ),
+    FunctionType.ARRAY: FunctionConfig(
+        output_purpose=Purpose.PROPERTY,
+        arg_count=InfiniteFunctionArgs,
+        output_type=ArrayType(type=DataType.STRING),
+    ),
+    FunctionType.LENGTH: FunctionConfig(
+        valid_inputs={DataType.STRING, DataType.ARRAY, DataType.MAP},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.INTEGER,
+        arg_count=1,
+    ),
+    FunctionType.SUM: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.FLOAT,
+            DataType.NUMBER,
+            DataType.NUMERIC,
+            DataType.BOOL,
+        },
+        output_purpose=Purpose.METRIC,
+        arg_count=1,
+    ),
+    FunctionType.ARRAY_AGG: FunctionConfig(
+        valid_inputs={*DataType},
+        output_purpose=Purpose.METRIC,
+        output_type_function=lambda args: ArrayType(
+            type=merge_datatypes([arg_to_datatype(x) for x in args])
+        ),
+        arg_count=1,
+    ),
+    FunctionType.ANY: FunctionConfig(
+        valid_inputs={*DataType},
+        output_purpose=Purpose.PROPERTY,
+        arg_count=1,
+    ),
+    FunctionType.BOOL_AND: FunctionConfig(
+        valid_inputs={DataType.BOOL},
+        output_purpose=Purpose.METRIC,
+        output_type=DataType.BOOL,
+        arg_count=1,
+    ),
+    FunctionType.BOOL_OR: FunctionConfig(
+        valid_inputs={DataType.BOOL},
+        output_purpose=Purpose.METRIC,
+        output_type=DataType.BOOL,
+        arg_count=1,
+    ),
+    FunctionType.AVG: FunctionConfig(
+        valid_inputs={
+            DataType.INTEGER,
+            DataType.FLOAT,
+            DataType.NUMBER,
+            DataType.NUMERIC,
+        },
+        output_purpose=Purpose.METRIC,
+        arg_count=1,
+    ),
+    FunctionType.UNIX_TO_TIMESTAMP: FunctionConfig(
+        valid_inputs={DataType.INTEGER},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.TIMESTAMP,
+        arg_count=1,
+    ),
+    FunctionType.HASH: FunctionConfig(
+        valid_inputs={
+            DataType.STRING,
+        },
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.STRING,
+        arg_count=2,
+    ),
+    FunctionType.GEO_POINT: FunctionConfig(
+        valid_inputs={DataType.NUMERIC},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.GEOGRAPHY,
+        arg_count=2,
+    ),
+    FunctionType.GEO_DISTANCE: FunctionConfig(
+        valid_inputs={DataType.GEOGRAPHY},
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.NUMERIC,
+        arg_count=2,
+    ),
+}
+
+EXCLUDED_FUNCTIONS = {
+    FunctionType.CUSTOM,
+    # Temporary
+    FunctionType.DATE_LITERAL,
+    FunctionType.DATETIME_LITERAL,
+    FunctionType.ARRAY,
+}
+
+for k in FunctionType.__members__.values():
+    if k not in FUNCTION_REGISTRY and k not in EXCLUDED_FUNCTIONS:
+        raise InvalidSyntaxException(
+            f"Function enum value {k} not in creation registry"
+        )
+
+
+class FunctionFactory:
+    def __init__(self, environment: Environment | None = None):
+        self.environment = environment
+
+    def create_function(
+        self,
+        args: list[Any],
+        operator: FunctionType,
+        meta: Meta | None = None,
+    ):
+        if operator not in FUNCTION_REGISTRY:
+            raise ValueError(f"Function {operator} not in registry")
+        config = FUNCTION_REGISTRY[operator]
+        valid_inputs: set[DataType] | list[set[DataType]] = config.valid_inputs or set(
+            DataType
+        )
+        output_purpose = config.output_purpose
+        base_output_type = config.output_type
+        arg_count = config.arg_count
+
+        if args:
+            if not self.environment:
+                raise ValueError("Environment required for function creation with args")
+            # TODO: remove this dependency
+            from trilogy.parsing.common import process_function_args
+
+            full_args = process_function_args(
+                args, environment=self.environment, meta=meta
+            )
+        else:
+            full_args = []
+        final_output_type: CONCRETE_TYPES
+        if config.output_type_function:
+
+            final_output_type = config.output_type_function(full_args)
+        elif not base_output_type:
+
+            final_output_type = merge_datatypes([arg_to_datatype(x) for x in full_args])
+        elif base_output_type:
+            final_output_type = base_output_type
+        else:
+            raise SyntaxError(f"Could not determine output type for {operator}")
+        if isinstance(final_output_type, TraitDataType) and self.environment:
+            final_output_type = TraitDataType(
+                type=final_output_type.type,
+                traits=[
+                    x
+                    for x in final_output_type.traits
+                    if operator
+                    not in self.environment.data_types.get(
+                        x, CUSTOM_PLACEHOLDER
+                    ).drop_on
+                ],
+            )
+
+        if not output_purpose:
+            if operator in FunctionClass.AGGREGATE_FUNCTIONS.value:
+                output_purpose = Purpose.METRIC
+            else:
+                output_purpose = Purpose.PROPERTY
+
+        return Function(
+            operator=operator,
+            arguments=full_args,  # type: ignore
+            output_datatype=final_output_type,
+            output_purpose=output_purpose,
+            valid_inputs=valid_inputs,
+            arg_count=arg_count,
+        )
+
+
+def create_function_derived_concept(
+    name: str,
+    namespace: str,
+    operator: FunctionType,
+    arguments: list[Concept],
+    environment: Environment,
+    output_type: Optional[
+        DataType | ArrayType | StructType | MapType | NumericType | TraitDataType
+    ] = None,
+    output_purpose: Optional[Purpose] = None,
+) -> Concept:
+    purpose = (
+        function_args_to_output_purpose(arguments, environment=environment)
+        if output_purpose is None
+        else output_purpose
+    )
+    output_type = arg_to_datatype(arguments[0]) if output_type is None else output_type
+    return Concept(
+        name=name,
+        namespace=namespace,
+        datatype=output_type,
+        purpose=purpose,
+        lineage=Function(
+            operator=operator,
+            arguments=[x.reference for x in arguments],
+            output_datatype=output_type,
+            output_purpose=purpose,
+            arg_count=len(arguments),
+        ),
+    )
+
+
+def argument_to_purpose(arg) -> Purpose:
+    if isinstance(arg, UndefinedConcept):
+        return Purpose.UNKNOWN
+    if isinstance(arg, Function):
+        return arg.output_purpose
+    elif isinstance(arg, AggregateWrapper):
+        base = arg.function.output_purpose
+        if arg.by and base == Purpose.METRIC:
+            return Purpose.PROPERTY
+        return arg.function.output_purpose
+    elif isinstance(arg, Parenthetical):
+        return argument_to_purpose(arg.content)
+    elif isinstance(arg, WindowItem):
+        return Purpose.PROPERTY
+    elif isinstance(arg, Conditional):
+        return Purpose.PROPERTY
+    elif isinstance(arg, Concept):
+        base = arg.purpose
+        if (
+            isinstance(arg.lineage, AggregateWrapper)
+            and arg.lineage.by
+            and base == Purpose.METRIC
+        ):
+            return Purpose.PROPERTY
+        return arg.purpose
+    elif isinstance(arg, (int, float, str, bool, list, NumericType, DataType)):
+        return Purpose.CONSTANT
+    elif isinstance(arg, DatePart):
+        return Purpose.CONSTANT
+    elif isinstance(arg, MagicConstants):
+        return Purpose.CONSTANT
+    else:
+        raise ValueError(f"Cannot parse arg purpose for {arg} of type {type(arg)}")
+
+
+def function_args_to_output_purpose(args, environment: Environment) -> Purpose:
+    has_metric = False
+    has_non_constant = False
+    has_non_single_row_constant = False
+    if not args:
+        return Purpose.CONSTANT
+    for arg in args:
+        if isinstance(arg, ConceptRef):
+            arg = environment.concepts[arg.address]
+        purpose = argument_to_purpose(arg)
+        if purpose == Purpose.METRIC:
+            has_metric = True
+        if purpose != Purpose.CONSTANT:
+            has_non_constant = True
+        if isinstance(arg, Concept) and arg.granularity != Granularity.SINGLE_ROW:
+            has_non_single_row_constant = True
+    if args and not has_non_constant and not has_non_single_row_constant:
+        return Purpose.CONSTANT
+    if has_metric:
+        return Purpose.METRIC
+    return Purpose.PROPERTY
+
+
+def Count(args: list[Concept], environment: Environment) -> Function:
+    return FunctionFactory(environment).create_function(
+        args=args, operator=FunctionType.COUNT
+    )
+
+
+def CountDistinct(args: list[Concept], environment: Environment) -> Function:
+    return FunctionFactory(environment).create_function(
+        args=args, operator=FunctionType.COUNT
+    )
+
+
+def Max(args: list[Concept], environment: Environment) -> Function:
+    return FunctionFactory(environment).create_function(
+        args=args, operator=FunctionType.COUNT
+    )
+
+
+def Min(args: list[Concept], environment: Environment) -> Function:
+    return FunctionFactory(environment).create_function(
+        args=args, operator=FunctionType.COUNT
+    )
+
+
+def Split(args: list[Concept], environment: Environment) -> Function:
+    return FunctionFactory(environment).create_function(
+        args=args, operator=FunctionType.SPLIT
+    )
+
+
+def AttrAccess(args: list[ConceptRef | str | int], environment: Environment):
+    return FunctionFactory(environment).create_function(
+        args=args, operator=FunctionType.ATTR_ACCESS
+    )
+
+
+def CurrentDate(
+    args: list[Concept], environment: Environment | None = None
+) -> Function:
+    return FunctionFactory(environment).create_function(
+        args=args, operator=FunctionType.CURRENT_DATE
+    )
+
+
+def CurrentDatetime(
+    args: list[Concept], environment: Environment | None = None
+) -> Function:
+    return FunctionFactory(environment).create_function(
+        args=args, operator=FunctionType.CURRENT_DATETIME
+    )