pytrilogy 0.0.3.85__py3-none-any.whl → 0.0.3.88__py3-none-any.whl

{pytrilogy-0.0.3.85.dist-info → pytrilogy-0.0.3.88.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pytrilogy
-Version: 0.0.3.85
+Version: 0.0.3.88
 Summary: Declarative, typed query language that compiles to SQL.
 Home-page: 
 Author: 

{pytrilogy-0.0.3.85.dist-info → pytrilogy-0.0.3.88.dist-info}/RECORD

@@ -1,5 +1,5 @@
-pytrilogy-0.0.3.85.dist-info/licenses/LICENSE.md,sha256=5ZRvtTyCCFwz1THxDTjAu3Lidds9WjPvvzgVwPSYNDo,1042
-trilogy/__init__.py,sha256=QKLkKDYevuT744TmiXXQu3mkG_tT9lB3iztr5_Mq7tU,303
+pytrilogy-0.0.3.88.dist-info/licenses/LICENSE.md,sha256=5ZRvtTyCCFwz1THxDTjAu3Lidds9WjPvvzgVwPSYNDo,1042
+trilogy/__init__.py,sha256=cl4nRewdhba3xQodaj4EqSuH_HCoF56PNkWEO79F_VY,303
 trilogy/compiler.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/constants.py,sha256=eKb_EJvSqjN9tGbdVEViwdtwwh8fZ3-jpOEDqL71y70,1691
 trilogy/engine.py,sha256=OK2RuqCIUId6yZ5hfF8J1nxGP0AJqHRZiafcowmW0xc,1728
@@ -11,25 +11,25 @@ trilogy/utility.py,sha256=euQccZLKoYBz0LNg5tzLlvv2YHvXh9HArnYp1V3uXsM,763
 trilogy/authoring/__init__.py,sha256=e74k-Jep4DLYPQU_2m0aVsYlw5HKTOucAKtdTbd6f2g,2595
 trilogy/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/core/constants.py,sha256=nizWYDCJQ1bigQMtkNIEMNTcN0NoEAXiIHLzpelxQ24,201
-trilogy/core/enums.py,sha256=gIbFV6HT93wUWgfOm6lQeMS_sQcvHc4IK-vuQpwyepY,8158
+trilogy/core/enums.py,sha256=RQRkpGHLtcBKAO6jZnmGVtSUnb00Q2rP56ltYGdfTok,8294
 trilogy/core/env_processor.py,sha256=pFsxnluKIusGKx1z7tTnfsd_xZcPy9pZDungkjkyvI0,3170
 trilogy/core/environment_helpers.py,sha256=VvPIiFemqaLLpIpLIqprfu63K7muZ1YzNg7UZIUph8w,8267
 trilogy/core/ergonomics.py,sha256=e-7gE29vPLFdg0_A1smQ7eOrUwKl5VYdxRSTddHweRA,1631
 trilogy/core/exceptions.py,sha256=jYEduuMehcMkmCpf-OC_taELPZm7qNfeSNzIWkDYScs,707
-trilogy/core/functions.py,sha256=hIvMYbSSsVDsXfHtZ7pwGc2CFIYqsVuYt9GOfAw3BRk,32013
+trilogy/core/functions.py,sha256=piMv91RiW1OAgDS5W8HIiO6fGleiNXIFW0YnNZgYOWU,32943
 trilogy/core/graph_models.py,sha256=BYhJzHKSgnZHVLJs1CfsgrxTPHqKqPNeA64RlozGY0A,3498
 trilogy/core/internal.py,sha256=wFx4e1I0mtx159YFShSXeUBSQ82NINtAbOI-92RX4i8,2151
 trilogy/core/optimization.py,sha256=ojpn-p79lr03SSVQbbw74iPCyoYpDYBmj1dbZ3oXCjI,8860
 trilogy/core/query_processor.py,sha256=5aFgv-2LVM1Uku9cR_tFuTRDwyLnxc95bCMAHeFy2AY,20332
 trilogy/core/utility.py,sha256=3VC13uSQWcZNghgt7Ot0ZTeEmNqs__cx122abVq9qhM,410
 trilogy/core/models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-trilogy/core/models/author.py,sha256=Xt_TRVl4xHF6fe-KTKPSt4Z8Srmplnv2KQe2grZXX3w,78411
-trilogy/core/models/build.py,sha256=0TRcL__hvajLcpFwYeLjCV35cow5oVPbJHXA3oO09XA,65931
+trilogy/core/models/author.py,sha256=5o0x9d3U_J2b8GPdJnMsMruC2P-xiI55lZKWUwRIeGU,78936
+trilogy/core/models/build.py,sha256=CyrSo4xgU-uDKW3xUVYs5cTk3Z3Z2BMWdGQNHnHZOqU,66127
 trilogy/core/models/build_environment.py,sha256=s_C9xAHuD3yZ26T15pWVBvoqvlp2LdZ8yjsv2_HdXLk,5363
-trilogy/core/models/core.py,sha256=q-8ExAaSB3PoSDX2XOqfZ01HBBvT1pZFs7jGg1vZizc,11096
+trilogy/core/models/core.py,sha256=eSzrrscwTrlO_PIn9_UxKSPBgKgR9-xp90LY1e0lhN0,11538
 trilogy/core/models/datasource.py,sha256=wogTevZ-9CyUW2a8gjzqMCieircxi-J5lkI7EOAZnck,9596
 trilogy/core/models/environment.py,sha256=0IHSCFf5e5b4LPQN3vmjumtfM1iD1tN4WMoUr0UqxZI,27855
-trilogy/core/models/execute.py,sha256=Erufst3yroU3PgNJMWF5lICm14AkH02laYeesSB01UU,41705
+trilogy/core/models/execute.py,sha256=sVWhjwWull-T6pUJizhrYVGCWHY3eZivVN6KNlhcHig,41839
 trilogy/core/optimizations/__init__.py,sha256=YH2-mGXZnVDnBcWVi8vTbrdw7Qs5TivG4h38rH3js_I,290
 trilogy/core/optimizations/base_optimization.py,sha256=gzDOKImoFn36k7XBD3ysEYDnbnb6vdVIztUfFQZsGnM,513
 trilogy/core/optimizations/inline_datasource.py,sha256=2sWNRpoRInnTgo9wExVT_r9RfLAQHI57reEV5cGHUcg,4329
@@ -77,15 +77,15 @@ trilogy/core/statements/build.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hS
 trilogy/core/statements/common.py,sha256=KxEmz2ySySyZ6CTPzn0fJl5NX2KOk1RPyuUSwWhnK1g,759
 trilogy/core/statements/execute.py,sha256=pfr1CZ_Cx1qQ-7LDyRI0JUfvtxBr_GGv-VeqiAjr43g,1406
 trilogy/dialect/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-trilogy/dialect/base.py,sha256=Bj3Wun81NAtfpjAyJGHDPSogPaTjBO1Bnmm9JxfpyfM,47533
+trilogy/dialect/base.py,sha256=bkCJsJvUQLlrKzHCTbWlLikas5o-JluCVI12RSSbYuE,48228
 trilogy/dialect/bigquery.py,sha256=8xhEu0z_lKANjbvzvBbC7CeKrJf1iP8YyrHqNale-ug,4351
 trilogy/dialect/common.py,sha256=tSthIZOXXRPQ4KeMKnDDsH7KlTmf2EVqigVtLyoc4zc,6071
 trilogy/dialect/config.py,sha256=olnyeVU5W5T6b9-dMeNAnvxuPlyc2uefb7FRME094Ec,3834
 trilogy/dialect/dataframe.py,sha256=RUbNgReEa9g3pL6H7fP9lPTrAij5pkqedpZ99D8_5AE,1522
-trilogy/dialect/duckdb.py,sha256=uqnZEd1IH4tO8FSbbUXe3Jl9EjoyrDde406k0VDuMXw,5544
+trilogy/dialect/duckdb.py,sha256=JoUvQ19WvgxoaJkGLM7DPXOd1H0394k3vBiblksQzOI,5631
 trilogy/dialect/enums.py,sha256=FRNYQ5-w-B6-X0yXKNU5g9GowsMlERFogTC5u2nxL_s,4740
 trilogy/dialect/postgres.py,sha256=el2PKwfyvWGk5EZtLudqAH5ewLitY1sFHJiocBSyxyM,3393
-trilogy/dialect/presto.py,sha256=yzSF8SZ6o1dizj1UueAa7S3lR0qNYJdSXbF78EHyhY0,3668
+trilogy/dialect/presto.py,sha256=k1IaeilR3nzPC9Hp7jlAdzJ7TsuxB3LQTBQ28MYE7O8,3715
 trilogy/dialect/snowflake.py,sha256=T6_mKfhpDazB1xQxqFLS2AJwzwzBcPYY6_qxRnAtFBs,3326
 trilogy/dialect/sql_server.py,sha256=HnUEvb8Yjl6MnMWTITzpFPZgDajhGzfDPz5A8dDerak,3279
 trilogy/hooks/__init__.py,sha256=T3SF3phuUDPLXKGRVE_Lf9mzuwoXWyaLolncR_1kY30,144
@@ -98,9 +98,9 @@ trilogy/parsing/common.py,sha256=yV1AckK0h8u1OFeGQBTMu-wuW5m63c5CcZuPicsTH_w,306
 trilogy/parsing/config.py,sha256=Z-DaefdKhPDmSXLgg5V4pebhSB0h590vI0_VtHnlukI,111
 trilogy/parsing/exceptions.py,sha256=Xwwsv2C9kSNv2q-HrrKC1f60JNHShXcCMzstTSEbiCw,154
 trilogy/parsing/helpers.py,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
-trilogy/parsing/parse_engine.py,sha256=MXG5nzQWd1tVaJddF_NDst96N_ayALQ0EB3JMh8kWaY,78776
+trilogy/parsing/parse_engine.py,sha256=snT7m7LWrOHcSZlqdJ5HXWmH1wMMPDaqrsf6AQsMx70,79800
 trilogy/parsing/render.py,sha256=HSNntD82GiiwHT-TWPLXAaIMWLYVV5B5zQEsgwrHIBE,19605
-trilogy/parsing/trilogy.lark,sha256=B7onP5W7ZIW_uKObJ5rL1sQCpnWiMeZ2azQ9j2U2ssE,15368
+trilogy/parsing/trilogy.lark,sha256=ySzMMLxyPjn74MjFHZxXPTW-jHW68KLPJpiszPvZaO0,15780
 trilogy/scripts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/scripts/trilogy.py,sha256=1L0XrH4mVHRt1C9T1HnaDv2_kYEfbWTb5_-cBBke79w,3774
 trilogy/std/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -111,8 +111,8 @@ trilogy/std/money.preql,sha256=XWwvAV3WxBsHX9zfptoYRnBigcfYwrYtBHXTME0xJuQ,2082
 trilogy/std/net.preql,sha256=WZCuvH87_rZntZiuGJMmBDMVKkdhTtxeHOkrXNwJ1EE,416
 trilogy/std/ranking.preql,sha256=LDoZrYyz4g3xsII9XwXfmstZD-_92i1Eox1UqkBIfi8,83
 trilogy/std/report.preql,sha256=LbV-XlHdfw0jgnQ8pV7acG95xrd1-p65fVpiIc-S7W4,202
-pytrilogy-0.0.3.85.dist-info/METADATA,sha256=FTQH7poVd-VpFg3tbI6vjCVJON0KYSxXUIdcS5CDHAc,9589
-pytrilogy-0.0.3.85.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-pytrilogy-0.0.3.85.dist-info/entry_points.txt,sha256=ewBPU2vLnVexZVnB-NrVj-p3E-4vukg83Zk8A55Wp2w,56
-pytrilogy-0.0.3.85.dist-info/top_level.txt,sha256=cAy__NW_eMAa_yT9UnUNlZLFfxcg6eimUAZ184cdNiE,8
-pytrilogy-0.0.3.85.dist-info/RECORD,,
+pytrilogy-0.0.3.88.dist-info/METADATA,sha256=MJ92RHQZuHjfV6eUuy9Z_F7U3BBt4ilDIzrfkS_kzSg,9589
+pytrilogy-0.0.3.88.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+pytrilogy-0.0.3.88.dist-info/entry_points.txt,sha256=ewBPU2vLnVexZVnB-NrVj-p3E-4vukg83Zk8A55Wp2w,56
+pytrilogy-0.0.3.88.dist-info/top_level.txt,sha256=cAy__NW_eMAa_yT9UnUNlZLFfxcg6eimUAZ184cdNiE,8
+pytrilogy-0.0.3.88.dist-info/RECORD,,

trilogy/__init__.py

@@ -4,6 +4,6 @@ from trilogy.dialect.enums import Dialects
 from trilogy.executor import Executor
 from trilogy.parser import parse
 
-__version__ = "0.0.3.85"
+__version__ = "0.0.3.88"
 
 __all__ = ["parse", "Executor", "Dialects", "Environment", "CONFIG"]

trilogy/core/enums.py

@@ -13,6 +13,11 @@ class UnnestMode(Enum):
     SNOWFLAKE = "snowflake"
 
 
+class GroupMode(Enum):
+    AUTO = "auto"
+    BY_INDEX = "by_index"
+
+
 class ConceptSource(Enum):
     MANUAL = "manual"
     CTE = "cte"
@@ -156,6 +161,10 @@ class FunctionType(Enum):
     ARRAY_TRANSFORM = "array_transform"
     ARRAY_TO_STRING = "array_to_string"
 
+    # MAP
+    MAP_KEYS = "map_keys"
+    MAP_VALUES = "map_values"
+
     # TEXT AND MAYBE MORE
     SPLIT = "split"
     LENGTH = "len"

trilogy/core/functions.py

@@ -70,7 +70,7 @@ def get_coalesce_output_type(args: list[Any]) -> CONCRETE_TYPES:
 
 
 def get_transform_output_type(args: list[Any]) -> CONCRETE_TYPES:
-    return arg_to_datatype(args[2])
+    return ArrayType(type=arg_to_datatype(args[2]))
 
 
 def get_index_output_type(
@@ -178,6 +178,20 @@ def get_date_trunc_output(
         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,
@@ -254,6 +268,22 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
         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.ARRAY_DISTINCT: FunctionConfig(
         valid_inputs={
             DataType.ARRAY,

trilogy/core/models/author.py

@@ -164,6 +164,21 @@ class ConceptRef(Addressable, Namespaced, DataTyped, Mergeable, BaseModel):
     def with_reference_replacement(self, source: str, target: Expr | ArgBinding):
         if self.address == source:
             return target
+
+        # a reference might be to an attribute of a struct that is bound late
+        # if the replacement is a parent in the access path; replace reference
+        # with an attribute access call
+        candidates = [f"local.{self.address}", self.address]
+        for candidate in candidates:
+            if not candidate.startswith(f"{source}."):
+                continue
+            return Function(
+                arguments=[target, self.address.rsplit(".", 1)[1]],
+                operator=FunctionType.ATTR_ACCESS,
+                arg_count=2,
+                output_datatype=arg_to_datatype(target),
+                output_purpose=Purpose.PROPERTY,
+            )
         return self
 
 
@@ -637,33 +652,31 @@ class Comparison(ConceptArgs, Mergeable, DataTyped, Namespaced, BaseModel):
 
     @model_validator(mode="after")
     def validate_comparison(self):
+        left_type = arg_to_datatype(self.left)
+        right_type = arg_to_datatype(self.right)
         if self.operator in (ComparisonOperator.IS, ComparisonOperator.IS_NOT):
-            if self.right != MagicConstants.NULL and DataType.BOOL != arg_to_datatype(
-                self.right
-            ):
+            if self.right != MagicConstants.NULL and DataType.BOOL != right_type:
                 raise SyntaxError(
                     f"Cannot use {self.operator.value} with non-null or boolean value {self.right}"
                 )
         elif self.operator in (ComparisonOperator.IN, ComparisonOperator.NOT_IN):
-            right_type = arg_to_datatype(self.right)
+
             if isinstance(right_type, ArrayType) and not is_compatible_datatype(
-                arg_to_datatype(self.left), right_type.value_data_type
+                left_type, right_type.value_data_type
             ):
                 raise SyntaxError(
-                    f"Cannot compare {arg_to_datatype(self.left)} and {right_type} with operator {self.operator} in {str(self)}"
+                    f"Cannot compare {left_type} and {right_type} with operator {self.operator} in {str(self)}"
                 )
             elif isinstance(self.right, Concept) and not is_compatible_datatype(
-                arg_to_datatype(self.left), arg_to_datatype(self.right)
+                left_type, right_type
             ):
                 raise SyntaxError(
-                    f"Cannot compare {arg_to_datatype(self.left)} and {arg_to_datatype(self.right)} with operator {self.operator} in {str(self)}"
+                    f"Cannot compare {left_type.name} and {right_type.name} with operator {self.operator} in {str(self)}"
                 )
         else:
-            if not is_compatible_datatype(
-                arg_to_datatype(self.left), arg_to_datatype(self.right)
-            ):
+            if not is_compatible_datatype(left_type, right_type):
                 raise SyntaxError(
-                    f"Cannot compare {arg_to_datatype(self.left)} and {arg_to_datatype(self.right)} of different types with operator {self.operator} in {str(self)}"
+                    f"Cannot compare {left_type.name} ({self.left}) and {right_type.name} ({self.right}) of different types with operator {self.operator.value} in {str(self)}"
                 )
 
         return self

trilogy/core/models/build.py

@@ -1649,8 +1649,11 @@ class Factory:
             full = self.local_concepts[base.address]
             if isinstance(full, BuildConcept):
                 return full
-        raw = self.environment.concepts[base.address]
-        return self.build(raw)
+        if base.address in self.environment.concepts:
+            raw = self.environment.concepts[base.address]
+            return self.build(raw)
+        # this will error by design - TODO - more helpful message?
+        return self.build(self.environment.concepts[base.address])
 
     @build.register
     def _(self, base: CaseWhen) -> BuildCaseWhen:

trilogy/core/models/core.py

@@ -27,7 +27,7 @@ from pydantic_core import core_schema
 from trilogy.constants import (
     MagicConstants,
 )
-from trilogy.core.enums import Ordering
+from trilogy.core.enums import Modifier, Ordering
 
 
 class DataTyped(ABC):
@@ -211,9 +211,19 @@ class MapType(BaseModel):
         return self.key_type
 
 
+class StructComponent(BaseModel):
+    name: str
+    type: TYPEDEF_TYPES
+    modifiers: list[Modifier] = []
+
+    @field_validator("type", mode="plain")
+    def validate_Type(cls, v):
+        return v
+
+
 class StructType(BaseModel):
-    fields: Sequence[TYPEDEF_TYPES]
-    fields_map: Dict[str, DataTyped | int | float | str]
+    fields: Sequence[StructComponent | TYPEDEF_TYPES]
+    fields_map: Dict[str, DataTyped | int | float | str | StructComponent]
 
     @field_validator("fields", mode="plain")
     def validate_type(cls, v):
@@ -234,6 +244,9 @@ class StructType(BaseModel):
     def value(self):
         return self.data_type.value
 
+    def __hash__(self):
+        return hash(str(self))
+
 
 class ListWrapper(Generic[VT], UserList):
     """Used to distinguish parsed list objects from other lists"""
@@ -401,6 +414,8 @@ def arg_to_datatype(arg) -> CONCRETE_TYPES:
         return DataType.STRING
     elif isinstance(arg, float):
         return DataType.FLOAT
+    elif isinstance(arg, DataType):
+        return arg
     elif isinstance(arg, NumericType):
         return arg
     elif isinstance(arg, TraitDataType):
@@ -420,6 +435,8 @@ def arg_to_datatype(arg) -> CONCRETE_TYPES:
         return DataType.DATETIME
     elif isinstance(arg, date):
         return DataType.DATE
+    elif isinstance(arg, StructComponent):
+        return arg_to_datatype(arg.type)
     else:
         raise ValueError(
             f"Cannot parse arg datatype for arg of raw type {type(arg)} value {arg}"

trilogy/core/models/execute.py

@@ -1076,6 +1076,11 @@ class UnionCTE(BaseModel):
     def group_to_grain(self) -> bool:
         return False
 
+    @property
+    def group_concepts(self) -> List[BuildConcept]:
+        # unions should always be on unique sets
+        return []
+
     def __add__(self, other):
         if not isinstance(other, UnionCTE) or not other.name == self.name:
             raise SyntaxError("Cannot merge union CTEs")

trilogy/dialect/base.py

@@ -15,6 +15,7 @@ from trilogy.core.enums import (
     ComparisonOperator,
     DatePart,
     FunctionType,
+    GroupMode,
     Ordering,
     ShowCategory,
     UnnestMode,
@@ -186,6 +187,9 @@ FUNCTION_MAP = {
     FunctionType.ARRAY: lambda x: f"[{', '.join(x)}]",
     FunctionType.DATE_LITERAL: lambda x: f"date '{x}'",
     FunctionType.DATETIME_LITERAL: lambda x: f"datetime '{x}'",
+    # MAP
+    FunctionType.MAP_KEYS: lambda x: f"map_keys({x[0]})",
+    FunctionType.MAP_VALUES: lambda x: f"map_values({x[0]})",
     # ARRAY
     FunctionType.ARRAY_SUM: lambda x: f"array_sum({x[0]})",
     FunctionType.ARRAY_DISTINCT: lambda x: f"array_distinct({x[0]})",
@@ -338,6 +342,7 @@ class BaseDialect:
     DATATYPE_MAP = DATATYPE_MAP
     COMPLEX_DATATYPE_MAP = COMPLEX_DATATYPE_MAP
     UNNEST_MODE = UnnestMode.CROSS_APPLY
+    GROUP_MODE = GroupMode.AUTO
 
     def __init__(self, rendering: Rendering | None = None):
         self.rendering = rendering or CONFIG.rendering
@@ -775,6 +780,33 @@ class BaseDialect:
         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 render_cte(self, cte: CTE | UnionCTE, auto_sort: bool = True) -> CompiledCTE:
         if isinstance(cte, UnionCTE):
             base_statement = f"\n{cte.operator}\n".join(
@@ -926,20 +958,7 @@ class BaseDialect:
                     if cte.order_by
                     else None
                 ),
-                group_by=(
-                    sorted(
-                        list(
-                            set(
-                                [
-                                    self.render_concept_sql(c, cte, alias=False)
-                                    for c in cte.group_concepts
-                                ]
-                            )
-                        )
-                    )
-                    if cte.group_to_grain
-                    else None
-                ),
+                group_by=self.render_cte_group_by(cte, select_columns),
             ),
         )
 

trilogy/dialect/duckdb.py

@@ -76,6 +76,7 @@ FUNCTION_MAP = {
     FunctionType.DATE: lambda x: f"cast({x[0]} as date)",
     FunctionType.DATE_TRUNCATE: lambda x: f"date_trunc('{x[1]}', {x[0]})",
     FunctionType.DATE_ADD: lambda x: f"date_add({x[0]}, {x[2]} * INTERVAL 1 {x[1]})",
+    FunctionType.DATE_SUB: lambda x: f"date_add({x[0]}, -{x[2]} * INTERVAL 1 {x[1]})",
     FunctionType.DATE_PART: lambda x: f"date_part('{x[1]}', {x[0]})",
     FunctionType.DATE_DIFF: lambda x: f"date_diff('{x[2]}', {x[0]}, {x[1]})",
     FunctionType.CONCAT: lambda x: f"({' || '.join(x)})",

trilogy/dialect/presto.py

@@ -2,7 +2,7 @@ from typing import Any, Callable, Mapping
 
 from jinja2 import Template
 
-from trilogy.core.enums import FunctionType, UnnestMode, WindowType
+from trilogy.core.enums import FunctionType, GroupMode, UnnestMode, WindowType
 from trilogy.core.models.core import DataType
 from trilogy.dialect.base import BaseDialect
 
@@ -94,6 +94,7 @@ class PrestoDialect(BaseDialect):
         DataType.STRING: "VARCHAR",
     }
     UNNEST_MODE = UnnestMode.PRESTO
+    GROUP_MODE = GroupMode.BY_INDEX
 
 
 class TrinoDialect(PrestoDialect):

trilogy/parsing/parse_engine.py

@@ -89,6 +89,7 @@ from trilogy.core.models.core import (
     MapType,
     MapWrapper,
     NumericType,
+    StructComponent,
     StructType,
     TraitDataType,
     TupleWrapper,
@@ -443,20 +444,39 @@ class ParseToObjects(Transformer):
     def MINUS(self, args) -> str:
         return "-"
 
+    @v_args(meta=True)
+    def struct_component(self, meta: Meta, args) -> StructComponent:
+        modifiers = []
+        for arg in args:
+            if isinstance(arg, Modifier):
+                modifiers.append(arg)
+        return StructComponent(name=args[0], type=args[1], modifiers=modifiers)
+
     @v_args(meta=True)
     def struct_type(self, meta: Meta, args) -> StructType:
         final: list[
-            DataType | MapType | ArrayType | NumericType | StructType | Concept
+            DataType
+            | MapType
+            | ArrayType
+            | NumericType
+            | StructType
+            | StructComponent
+            | Concept
         ] = []
         for arg in args:
-            new = self.environment.concepts.__getitem__(  # type: ignore
-                key=arg, line_no=meta.line
-            )
-            final.append(new)
+            if isinstance(arg, StructComponent):
+                final.append(arg)
+            else:
+                new = self.environment.concepts.__getitem__(  # type: ignore
+                    key=arg, line_no=meta.line
+                )
+                final.append(new)
 
         return StructType(
             fields=final,
-            fields_map={x.name: x for x in final if isinstance(x, Concept)},
+            fields_map={
+                x.name: x for x in final if isinstance(x, (Concept, StructComponent))
+            },
         )
 
     def list_type(self, args) -> ArrayType:
@@ -683,7 +703,9 @@ class ParseToObjects(Transformer):
             )
             # let constant purposes exist to support round-tripping
             # as a build concept may end up with a constant based on constant inlining happening recursively
-            if (
+            if purpose == Purpose.KEY and concept.purpose != Purpose.KEY:
+                concept.purpose = Purpose.KEY
+            elif (
                 purpose
                 and purpose != Purpose.AUTO
                 and concept.purpose != purpose
@@ -2024,6 +2046,16 @@ class ParseToObjects(Transformer):
     def fbool(self, meta, args):
         return self.function_factory.create_function(args, FunctionType.BOOL, meta)
 
+    @v_args(meta=True)
+    def fmap_keys(self, meta, args):
+        return self.function_factory.create_function(args, FunctionType.MAP_KEYS, meta)
+
+    @v_args(meta=True)
+    def fmap_values(self, meta, args):
+        return self.function_factory.create_function(
+            args, FunctionType.MAP_VALUES, meta
+        )
+
     @v_args(meta=True)
     def farray_sum(self, meta, args):
         return self.function_factory.create_function(args, FunctionType.ARRAY_SUM, meta)

trilogy/parsing/trilogy.lark

@@ -181,7 +181,8 @@
     //indexing into an expression is a function
     index_access: atom "[" int_lit "]"
     map_key_access: atom "[" string_lit "]"
-    attr_access: atom  "." string_lit 
+    _ATTR_ACCESS.1 : "getattr("i
+    attr_access: ( _ATTR_ACCESS atom "," string_lit ")") | (atom  "." string_lit)
 
     ?expr: comparison_root | between_root
 
@@ -309,6 +310,14 @@
     farray_transform: _ARRAY_TRANSFORM expr ","  transform_lambda   ")"
 
     _array_functions: farray_sum | farray_distinct | farray_sort | farray_transform | farray_to_string
+
+    //map_functions
+    _MAP_KEYS.1: "map_keys("i
+    fmap_keys: _MAP_KEYS expr ")"
+    _MAP_VALUES.1: "map_values("i
+    fmap_values: _MAP_VALUES expr ")"
+
+    _map_functions: fmap_keys | fmap_values
     
     // special aggregate
     _GROUP.1: "group("i
@@ -367,7 +376,7 @@
     fyear: _YEAR expr ")"
     
     DATE_PART: "DAY"i | "WEEK"i | "MONTH"i | "QUARTER"i | "YEAR"i | "MINUTE"i | "HOUR"i | "SECOND"i | "DAY_OF_WEEK"i
-    _DATE_TRUNC.1: "date_trunc("i
+    _DATE_TRUNC.1: "date_trunc("i | "date_truncate("i
     fdate_trunc: _DATE_TRUNC expr "," DATE_PART ")"
     _DATE_PART.1: "date_part("i
     fdate_part: _DATE_PART expr "," DATE_PART ")"
@@ -380,7 +389,7 @@
     
     _date_functions: fdate | fdate_add | fdate_sub | fdate_diff | fdatetime | ftimestamp | fsecond | fminute | fhour | fday | fday_of_week | fweek | fmonth | fquarter | fyear | fdate_part | fdate_trunc
     
-    _static_functions:  _string_functions | _math_functions | _array_functions
+    _static_functions:  _string_functions | _math_functions | _array_functions | _map_functions
 
     custom_function:  "@" IDENTIFIER  "(" (expr ",")* expr ")"
 
@@ -424,8 +433,8 @@
     MODIFIER: /OPTIONAL|PARTIAL|NULLABLE/i
 
     SHORTHAND_MODIFIER: /~|\?/
-
-    struct_type: "struct"i "<" ((data_type | IDENTIFIER) ",")* (data_type | IDENTIFIER) ","? ">" 
+    struct_component: IDENTIFIER ":" data_type concept_nullable_modifier? metadata?
+    struct_type: "struct"i "<" ((struct_component | IDENTIFIER) ",")* (struct_component | IDENTIFIER) ","? ">" 
 
     list_type: ("list"i | "array"i) "<" (data_type | IDENTIFIER) ">"