pytrilogy 0.0.3.16__py3-none-any.whl → 0.0.3.18__py3-none-any.whl

{pytrilogy-0.0.3.16.dist-info → pytrilogy-0.0.3.18.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: pytrilogy
-Version: 0.0.3.16
+Version: 0.0.3.18
 Summary: Declarative, typed query language that compiles to SQL.
 Home-page: 
 Author: 

{pytrilogy-0.0.3.16.dist-info → pytrilogy-0.0.3.18.dist-info}/RECORD

@@ -1,4 +1,4 @@
-trilogy/__init__.py,sha256=4fwdD6idpRmJ7XtHqlcxrKzhBPI4tVYCM-NldzFMJIY,303
+trilogy/__init__.py,sha256=jrbWIot3RjwpdU7heWZocVfCD4BBtWw40jekbkQLvdg,303
 trilogy/compiler.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/constants.py,sha256=qZ1d0hoKPPV2HHCoFwPYTVB7b6bXjpWvXd3lE-zEhy8,1494
 trilogy/engine.py,sha256=OK2RuqCIUId6yZ5hfF8J1nxGP0AJqHRZiafcowmW0xc,1728
@@ -7,26 +7,26 @@ trilogy/parser.py,sha256=o4cfk3j3yhUFoiDKq9ZX_GjBF3dKhDjXEwb63rcBkBM,293
 trilogy/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/render.py,sha256=D6rI1RNtn0StJeSe4e18lnlc-U--cNu4lh5C_NkU_uM,1218
 trilogy/utility.py,sha256=euQccZLKoYBz0LNg5tzLlvv2YHvXh9HArnYp1V3uXsM,763
-trilogy/authoring/__init__.py,sha256=6KPn4uoPb_7t-i510M8UNXB7nrPpaKlxAXwuRbWBdLE,2115
+trilogy/authoring/__init__.py,sha256=ohkYA3_LGYZh3fwzEYKTN6ofACDI5GYl3VCbGxVvlzY,2233
 trilogy/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/core/constants.py,sha256=7XaCpZn5mQmjTobbeBn56SzPWq9eMNDfzfsRU-fP0VE,171
-trilogy/core/enums.py,sha256=jOgUyoLGgC5-8SLBWP0jQDMxptBJbN0J48VsCa1EMYo,7111
+trilogy/core/enums.py,sha256=ND69oja7DOsZS2T8JlIuDW2-uKm74x9SJOWbAqNeopU,7137
 trilogy/core/env_processor.py,sha256=pFsxnluKIusGKx1z7tTnfsd_xZcPy9pZDungkjkyvI0,3170
 trilogy/core/environment_helpers.py,sha256=oOpewPwMp8xOtx2ayeeyuLNUwr-cli7UanHKot5ebNY,7627
-trilogy/core/ergonomics.py,sha256=ASLDd0RqKWrZiG3XcKHo8nyTjaB_8xfE9t4NZ1UvGpc,1639
+trilogy/core/ergonomics.py,sha256=e-7gE29vPLFdg0_A1smQ7eOrUwKl5VYdxRSTddHweRA,1631
 trilogy/core/exceptions.py,sha256=JPYyBcit3T_pRtlHdtKSeVJkIyWUTozW2aaut25A2xI,673
-trilogy/core/functions.py,sha256=EsRYHE2kg_FckceVgYGPPs1ylrXvBjr3l1Wa6r0SvL8,25027
+trilogy/core/functions.py,sha256=E-OZZTshUNmEMMl8nhaUDUhQkr45gqUCdPaDiBlayic,25392
 trilogy/core/graph_models.py,sha256=z17EoO8oky2QOuO6E2aMWoVNKEVJFhLdsQZOhC4fNLU,2079
 trilogy/core/internal.py,sha256=iicDBlC6nM8d7e7jqzf_ZOmpUsW8yrr2AA8AqEiLx-s,1577
 trilogy/core/optimization.py,sha256=xGO8piVsLrpqrx-Aid_Y56_5slSv4eZmlP64hCHRiEc,7957
 trilogy/core/query_processor.py,sha256=Do8YpdPBdsbKtl9n37hobzk8SORMGqH-e_zNNxd-BE4,19456
 trilogy/core/models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-trilogy/core/models/author.py,sha256=CHVnj-GW_gjGDHaAUg60VGOfPSRr_5n5LbHvocOhQj8,70371
+trilogy/core/models/author.py,sha256=u2wS2rg3Qsqb5xNRUYvjgA7jOwQS2dMZzytlwJtglBs,70859
 trilogy/core/models/build.py,sha256=bO1qYvuGl6LeNGgsfS6ZHAzZBR2lBPLg-QJymp9hgkU,57235
 trilogy/core/models/build_environment.py,sha256=8UggvlPU708GZWYPJMc_ou2r7M3TY2g69eqGvz03YX0,5528
-trilogy/core/models/core.py,sha256=00opIUXwgJy9OF-cwI883zQpArNeh6wkTpSqUszDU78,9909
+trilogy/core/models/core.py,sha256=nb4h1HHm5_qwmUkYth4zRhEttS1EtsMZCP4vT20EEAE,10326
 trilogy/core/models/datasource.py,sha256=6RjJUd2u4nYmEwFBpJlM9LbHVYDv8iHJxqiBMZqUrwI,9422
-trilogy/core/models/environment.py,sha256=WCVJNWUze37bxFSFWo_6Z7330OG5k0DDmb7a2Tz5gg4,26264
+trilogy/core/models/environment.py,sha256=4w1a9iHdRuOuSoZxz5Kzoy61Vjj4TZVXzKkCSP6c-Ow,26298
 trilogy/core/models/execute.py,sha256=kxkw14vgUudHHELXLo73AHiaMEpLfiv0qAvbxjJxn_k,33929
 trilogy/core/optimizations/__init__.py,sha256=EBanqTXEzf1ZEYjAneIWoIcxtMDite5-n2dQ5xcfUtg,356
 trilogy/core/optimizations/base_optimization.py,sha256=gzDOKImoFn36k7XBD3ysEYDnbnb6vdVIztUfFQZsGnM,513
@@ -64,7 +64,7 @@ trilogy/core/processing/nodes/union_node.py,sha256=fDFzLAUh5876X6_NM7nkhoMvHEdGJ
 trilogy/core/processing/nodes/unnest_node.py,sha256=oLKMMNMx6PLDPlt2V5neFMFrFWxET8r6XZElAhSNkO0,2181
 trilogy/core/processing/nodes/window_node.py,sha256=STvwheVttxSWVHB-yUQUSo-Pyz7Uk8G1txFDAbWMp-s,1380
 trilogy/core/statements/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-trilogy/core/statements/author.py,sha256=X3NdGlWTUCNdQQw9EGI1kccgsLJMazwNAIsqpUSr_ZY,14443
+trilogy/core/statements/author.py,sha256=1cJ4nIBInTD3YcS89yTBda2ypghIhMvb-Mut66m_sxM,14615
 trilogy/core/statements/build.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/core/statements/common.py,sha256=KxEmz2ySySyZ6CTPzn0fJl5NX2KOk1RPyuUSwWhnK1g,759
 trilogy/core/statements/execute.py,sha256=cSlvpHFOqpiZ89pPZ5GDp9Hu6j6uj-5_h21FWm_L-KM,1248
@@ -74,7 +74,7 @@ trilogy/dialect/bigquery.py,sha256=PkjFcNGZHYOe655PmJhb8a0afdFULuovqP0qQVO8m0I,2
 trilogy/dialect/common.py,sha256=vYb-QPf_CnZ3mMLpOVjteWeLH1iaq2mn4WPx0XGoo20,4033
 trilogy/dialect/config.py,sha256=EGYRQIbrkeMuud5Bkds7jSD5dCJR5hEYZUYcy-lYZl4,3308
 trilogy/dialect/dataframe.py,sha256=RUbNgReEa9g3pL6H7fP9lPTrAij5pkqedpZ99D8_5AE,1522
-trilogy/dialect/duckdb.py,sha256=AXpvheZQ_hCzTgTRAM5cxBZwr1N4XFzEGwRC1CGPxt4,3705
+trilogy/dialect/duckdb.py,sha256=TepCOhYWYw1oUuOT6ZGlB3l4X6S8rYcldWe3zZm3HoU,3710
 trilogy/dialect/enums.py,sha256=QYIcVr5RgpYMA1Wl0nWeojVVxJxy0V2_sn8uqSFNx20,4615
 trilogy/dialect/postgres.py,sha256=VH4EB4myjIeZTHeFU6vK00GxY9c53rCBjg2mLbdaCEE,3254
 trilogy/dialect/presto.py,sha256=Mw7_F8h19mWfuZHkHQJizQWbpu1lIHe6t2PA0r88gsY,3392
@@ -86,18 +86,18 @@ trilogy/hooks/graph_hook.py,sha256=c-vC-IXoJ_jDmKQjxQyIxyXPOuUcLIURB573gCsAfzQ,2
 trilogy/hooks/query_debugger.py,sha256=1npRjww94sPV5RRBBlLqMJRaFkH9vhEY6o828MeoEcw,5583
 trilogy/metadata/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/parsing/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-trilogy/parsing/common.py,sha256=IgZ2K3LzJ0usLIwxRCRmS-4luP6uwmM-f1oqGNyGbm0,21306
+trilogy/parsing/common.py,sha256=99tDKrpQTk-SpyTXUqKFtm-lfmhjCOQIn25hxbQvRRg,21459
 trilogy/parsing/config.py,sha256=Z-DaefdKhPDmSXLgg5V4pebhSB0h590vI0_VtHnlukI,111
 trilogy/parsing/exceptions.py,sha256=92E5i2frv5hj9wxObJZsZqj5T6bglvPzvdvco_vW1Zk,38
 trilogy/parsing/helpers.py,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
-trilogy/parsing/parse_engine.py,sha256=YjYe4cAxek3CceewGI1It-XEO_ZygCxY9IQzcJw0Y0Y,57555
+trilogy/parsing/parse_engine.py,sha256=Hodpr2QjJNHSS5xGP-wPFjKK9gpygGG6loCikyRiwH8,60287
 trilogy/parsing/render.py,sha256=o_XuQWhcwx1lD9eGVqkqZEwkmQK0HdmWWokGBtdeH4I,17837
-trilogy/parsing/trilogy.lark,sha256=h7q1UBTBBhGbDTPWQ1COtf-Y41ZRHMP_B9iJ0WxLSfY,12831
+trilogy/parsing/trilogy.lark,sha256=7libFS5HNiyHJYzr5_lEiY-Lpqit04_PgyIPHMZT7-w,12928
 trilogy/scripts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/scripts/trilogy.py,sha256=1L0XrH4mVHRt1C9T1HnaDv2_kYEfbWTb5_-cBBke79w,3774
-pytrilogy-0.0.3.16.dist-info/LICENSE.md,sha256=5ZRvtTyCCFwz1THxDTjAu3Lidds9WjPvvzgVwPSYNDo,1042
-pytrilogy-0.0.3.16.dist-info/METADATA,sha256=bEP9eVWQb7cplZ0EeB5fBluzVoMy_jLd0W1X3ZmrGrs,8984
-pytrilogy-0.0.3.16.dist-info/WHEEL,sha256=jB7zZ3N9hIM9adW7qlTAyycLYW9npaWKLRzaoVcLKcM,91
-pytrilogy-0.0.3.16.dist-info/entry_points.txt,sha256=0petKryjvvtEfTlbZC1AuMFumH_WQ9v8A19LvoS6G6c,54
-pytrilogy-0.0.3.16.dist-info/top_level.txt,sha256=cAy__NW_eMAa_yT9UnUNlZLFfxcg6eimUAZ184cdNiE,8
-pytrilogy-0.0.3.16.dist-info/RECORD,,
+pytrilogy-0.0.3.18.dist-info/LICENSE.md,sha256=5ZRvtTyCCFwz1THxDTjAu3Lidds9WjPvvzgVwPSYNDo,1042
+pytrilogy-0.0.3.18.dist-info/METADATA,sha256=V6EpL5Z-9Xief5rqKaourYFLzmRMZH9jM31orwdgYAk,8984
+pytrilogy-0.0.3.18.dist-info/WHEEL,sha256=52BFRY2Up02UkjOa29eZOS2VxUrpPORXg1pkohGGUS8,91
+pytrilogy-0.0.3.18.dist-info/entry_points.txt,sha256=0petKryjvvtEfTlbZC1AuMFumH_WQ9v8A19LvoS6G6c,54
+pytrilogy-0.0.3.18.dist-info/top_level.txt,sha256=cAy__NW_eMAa_yT9UnUNlZLFfxcg6eimUAZ184cdNiE,8
+pytrilogy-0.0.3.18.dist-info/RECORD,,

{pytrilogy-0.0.3.16.dist-info → pytrilogy-0.0.3.18.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.8.2)
+Generator: setuptools (76.0.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

trilogy/__init__.py

@@ -4,6 +4,6 @@ from trilogy.dialect.enums import Dialects
 from trilogy.executor import Executor
 from trilogy.parser import parse
 
-__version__ = "0.0.3.16"
+__version__ = "0.0.3.18"
 
 __all__ = ["parse", "Executor", "Dialects", "Environment", "CONFIG"]

trilogy/authoring/__init__.py

@@ -40,6 +40,7 @@ from trilogy.core.models.core import (
     MapType,
     StructType,
 )
+from trilogy.core.models.datasource import Datasource, DatasourceMetadata
 from trilogy.core.models.environment import Environment
 from trilogy.core.statements.author import (
     ConceptDeclarationStatement,
@@ -100,4 +101,6 @@ __all__ = [
     "MultiSelectStatement",
     "PersistStatement",
     "RawSQLStatement",
+    "Datasource",
+    "DatasourceMetadata",
 ]

trilogy/core/enums.py

@@ -193,6 +193,7 @@ class FunctionType(Enum):
     DATE_PART = "date_part"
     DATE_TRUNCATE = "date_truncate"
     DATE_ADD = "date_add"
+    DATE_SUB = "date_sub"
     DATE_DIFF = "date_diff"
 
     # UNIX

trilogy/core/ergonomics.py

@@ -3,7 +3,6 @@ from trilogy.constants import CONFIG
 # source: https://github.com/aaronbassett/Pass-phrase
 CTE_NAMES = """quizzical
 highfalutin
-dynamic
 wakeful
 cheerful
 thoughtful

trilogy/core/functions.py

@@ -29,6 +29,7 @@ from trilogy.core.models.core import (
     MapType,
     NumericType,
     StructType,
+    TraitDataType,
     arg_to_datatype,
     merge_datatypes,
 )
@@ -94,6 +95,10 @@ def get_cast_output_type(
     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:
@@ -140,6 +145,12 @@ def get_date_trunc_output(
 
 
 FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
+    FunctionType.ALIAS: FunctionConfig(
+        arg_count=1,
+    ),
+    FunctionType.PARENTHETICAL: FunctionConfig(
+        arg_count=1,
+    ),
     FunctionType.UNNEST: FunctionConfig(
         valid_inputs={
             DataType.ARRAY,
@@ -173,7 +184,6 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
             DataType.BOOL,
         },
         output_purpose=Purpose.METRIC,
-        output_type=DataType.INTEGER,
         arg_count=1,
     ),
     FunctionType.MIN: FunctionConfig(
@@ -186,7 +196,6 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
             DataType.TIMESTAMP,
         },
         output_purpose=Purpose.METRIC,
-        output_type=DataType.INTEGER,
         arg_count=1,
     ),
     FunctionType.SPLIT: FunctionConfig(
@@ -237,7 +246,6 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
     FunctionType.ABS: FunctionConfig(
         valid_inputs={DataType.INTEGER, DataType.FLOAT, DataType.NUMBER},
         output_purpose=Purpose.PROPERTY,
-        output_type=DataType.INTEGER,
         arg_count=1,
     ),
     FunctionType.COALESCE: FunctionConfig(
@@ -370,6 +378,21 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
         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=[
             {
@@ -519,7 +542,6 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
             DataType.NUMERIC,
         },
         output_purpose=Purpose.PROPERTY,
-        output_type=DataType.INTEGER,
         arg_count=InfiniteFunctionArgs,
     ),
     FunctionType.SUBTRACT: FunctionConfig(
@@ -530,7 +552,6 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
             DataType.NUMERIC,
         },
         output_purpose=Purpose.PROPERTY,
-        output_type=DataType.INTEGER,
         arg_count=InfiniteFunctionArgs,
     ),
     FunctionType.MULTIPLY: FunctionConfig(
@@ -541,7 +562,6 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
             DataType.NUMERIC,
         },
         output_purpose=Purpose.PROPERTY,
-        output_type=DataType.INTEGER,
         arg_count=InfiniteFunctionArgs,
     ),
     FunctionType.DIVIDE: FunctionConfig(
@@ -552,7 +572,6 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
             DataType.NUMERIC,
         },
         output_purpose=Purpose.PROPERTY,
-        output_type=DataType.INTEGER,
         arg_count=InfiniteFunctionArgs,
     ),
     FunctionType.MOD: FunctionConfig(
@@ -570,7 +589,7 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
             {DataType.INTEGER},
         ],
         output_purpose=Purpose.PROPERTY,
-        output_type=DataType.INTEGER,
+        output_type_function=lambda args: get_output_type_at_index(args, 0),
         arg_count=2,
     ),
     FunctionType.CUSTOM: FunctionConfig(
@@ -621,13 +640,11 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
     FunctionType.SUM: FunctionConfig(
         valid_inputs={DataType.INTEGER, DataType.FLOAT, DataType.NUMBER},
         output_purpose=Purpose.METRIC,
-        output_type=DataType.INTEGER,
         arg_count=1,
     ),
     FunctionType.AVG: FunctionConfig(
         valid_inputs={DataType.INTEGER, DataType.FLOAT, DataType.NUMBER},
         output_purpose=Purpose.METRIC,
-        output_type=DataType.INTEGER,
         arg_count=1,
     ),
     FunctionType.UNIX_TO_TIMESTAMP: FunctionConfig(
@@ -640,8 +657,6 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
 
 EXCLUDED_FUNCTIONS = {
     FunctionType.CUSTOM,
-    FunctionType.ALIAS,
-    FunctionType.PARENTHETICAL,
     # Temporary
     FunctionType.DATE_LITERAL,
     FunctionType.DATETIME_LITERAL,
@@ -650,7 +665,9 @@ EXCLUDED_FUNCTIONS = {
 
 for k in FunctionType.__members__.values():
     if k not in FUNCTION_REGISTRY and k not in EXCLUDED_FUNCTIONS:
-        raise InvalidSyntaxException(f"Function {k} not in registry")
+        raise InvalidSyntaxException(
+            f"Function enum value {k} not in creation registry"
+        )
 
 
 class FunctionFactory:
@@ -686,8 +703,10 @@ class FunctionFactory:
             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
@@ -715,7 +734,7 @@ def create_function_derived_concept(
     operator: FunctionType,
     arguments: list[Concept],
     output_type: Optional[
-        DataType | ListType | StructType | MapType | NumericType
+        DataType | ListType | StructType | MapType | NumericType | TraitDataType
     ] = None,
     output_purpose: Optional[Purpose] = None,
 ) -> Concept:

trilogy/core/models/author.py

@@ -793,6 +793,10 @@ class Concept(Addressable, DataTyped, ConceptArgs, Mergeable, Namespaced, BaseMo
         base = f"{self.address}@{self.grain}"
         return base
 
+    @property
+    def is_internal(self) -> bool:
+        return self.namespace.startswith("_") or self.name.startswith("_")
+
     @property
     def reference(self) -> ConceptRef:
         return ConceptRef.model_construct(
@@ -1370,7 +1374,7 @@ class WindowItem(DataTyped, ConceptArgs, Mergeable, Namespaced, BaseModel):
 
 
 def get_basic_type(
-    type: DataType | ListType | StructType | MapType | NumericType,
+    type: DataType | ListType | StructType | MapType | NumericType | TraitDataType,
 ) -> DataType:
     if isinstance(type, ListType):
         return DataType.LIST
@@ -1380,6 +1384,8 @@ def get_basic_type(
         return DataType.MAP
     if isinstance(type, NumericType):
         return DataType.NUMERIC
+    if isinstance(type, TraitDataType):
+        return type.type
     return type
 
 
@@ -1531,7 +1537,9 @@ def get_concept_arguments(expr) -> List["ConceptRef"]:
 class Function(DataTyped, ConceptArgs, Mergeable, Namespaced, BaseModel):
     operator: FunctionType
     arg_count: int = Field(default=1)
-    output_datatype: DataType | ListType | StructType | MapType | NumericType
+    output_datatype: (
+        DataType | ListType | StructType | MapType | NumericType | TraitDataType
+    )
     output_purpose: Purpose
     valid_inputs: Optional[
         Union[
@@ -1629,23 +1637,30 @@ class Function(DataTyped, ConceptArgs, Mergeable, Namespaced, BaseModel):
         return v
 
     def with_reference_replacement(self, source: str, target: Expr):
+        from trilogy.core.functions import arg_to_datatype, merge_datatypes
+
+        nargs = [
+            (
+                c.with_reference_replacement(
+                    source,
+                    target,
+                )
+                if isinstance(
+                    c,
+                    Mergeable,
+                )
+                else c
+            )
+            for c in self.arguments
+        ]
+        if self.output_datatype == DataType.UNKNOWN:
+            new_output = merge_datatypes([arg_to_datatype(x) for x in nargs])
+        else:
+            new_output = self.output_datatype
         return Function.model_construct(
             operator=self.operator,
-            arguments=[
-                (
-                    c.with_reference_replacement(
-                        source,
-                        target,
-                    )
-                    if isinstance(
-                        c,
-                        Mergeable,
-                    )
-                    else c
-                )
-                for c in self.arguments
-            ],
-            output_datatype=self.output_datatype,
+            arguments=nargs,
+            output_datatype=new_output,
             output_purpose=self.output_purpose,
             valid_inputs=self.valid_inputs,
             arg_count=self.arg_count,
@@ -1777,7 +1792,7 @@ class AggregateWrapper(Mergeable, DataTyped, ConceptArgs, Namespaced, BaseModel)
         )
 
 
-class FilterItem(Namespaced, ConceptArgs, BaseModel):
+class FilterItem(DataTyped, Namespaced, ConceptArgs, BaseModel):
     content: ConceptRef
     where: "WhereClause"
 
@@ -1837,7 +1852,7 @@ class RowsetLineage(Namespaced, Mergeable, BaseModel):
         )
 
 
-class RowsetItem(Mergeable, ConceptArgs, Namespaced, BaseModel):
+class RowsetItem(Mergeable, DataTyped, ConceptArgs, Namespaced, BaseModel):
     content: ConceptRef
     rowset: RowsetLineage
 

trilogy/core/models/core.py

@@ -51,7 +51,13 @@ class Addressable(ABC):
 
 
 TYPEDEF_TYPES = Union[
-    "DataType", "MapType", "ListType", "NumericType", "StructType", "DataTyped"
+    "DataType",
+    "MapType",
+    "ListType",
+    "NumericType",
+    "StructType",
+    "DataTyped",
+    "TraitDataType",
 ]
 
 CONCRETE_TYPES = Union[
@@ -60,6 +66,7 @@ CONCRETE_TYPES = Union[
     "ListType",
     "NumericType",
     "StructType",
+    "TraitDataType",
 ]
 
 KT = TypeVar("KT")
@@ -101,6 +108,16 @@ class TraitDataType(BaseModel):
     type: DataType
     traits: list[str]
 
+    def __hash__(self):
+        return hash(self.type)
+
+    def __eq__(self, other):
+        if isinstance(other, DataType):
+            return self.type == other
+        elif isinstance(other, TraitDataType):
+            return self.type == other.type and self.traits == other.traits
+        return False
+
     @property
     def data_type(self):
         return self.type
@@ -312,8 +329,10 @@ def dict_to_map_wrapper(arg):
 
 
 def merge_datatypes(
-    inputs: list[DataType | ListType | StructType | MapType | NumericType],
-) -> DataType | ListType | StructType | MapType | NumericType:
+    inputs: list[
+        DataType | ListType | StructType | MapType | NumericType | TraitDataType
+    ],
+) -> DataType | ListType | StructType | MapType | NumericType | TraitDataType:
     """This is a temporary hack for doing between
     allowable datatype transformation matrix"""
     if len(inputs) == 1:
@@ -348,6 +367,7 @@ def is_compatible_datatype(left, right):
 
 
 def arg_to_datatype(arg) -> CONCRETE_TYPES:
+
     if isinstance(arg, MagicConstants):
         if arg == MagicConstants.NULL:
             return DataType.NULL

trilogy/core/models/environment.py

@@ -58,6 +58,7 @@ if TYPE_CHECKING:
 class Import:
     alias: str
     path: Path
+    input_path: str | None = None
 
 
 class BaseImportResolver(BaseModel):

trilogy/core/statements/author.py

@@ -134,7 +134,6 @@ class SelectStatement(HasUUID, SelectTypeMixin, BaseModel):
         output.grain = output.calculate_grain(environment)
 
         for x in selection:
-
             if x.is_undefined and environment.concepts.fail_on_missing:
                 environment.concepts.raise_undefined(
                     x.concept.address, meta.line_number if meta else None
@@ -158,7 +157,6 @@ class SelectStatement(HasUUID, SelectTypeMixin, BaseModel):
                 output.local_concepts[x.content.address] = environment.concepts[
                     x.content.address
                 ]  # .set_select_grain(output.grain, environment)
-
         output.validate_syntax(environment)
         return output
 
@@ -384,7 +382,12 @@ class MergeStatementV2(HasUUID, BaseModel):
 
 
 class ImportStatement(HasUUID, BaseModel):
+    # import abc.def as bar
+    # the bit after 'as', eg bar
     alias: str
+    # the bit after import, abc.def
+    input_path: str
+    # what it actually resolves to, typically a filepath
     path: Path
 
 

trilogy/dialect/duckdb.py

@@ -8,6 +8,7 @@ from trilogy.dialect.base import BaseDialect
 
 WINDOW_FUNCTION_MAP: Mapping[WindowType, Callable[[Any, Any, Any], str]] = {}
 
+
 FUNCTION_MAP = {
     FunctionType.COUNT: lambda args: f"count({args[0]})",
     FunctionType.SUM: lambda args: f"sum({args[0]})",
@@ -27,7 +28,7 @@ FUNCTION_MAP = {
     # datetime is aliased
     FunctionType.CURRENT_DATETIME: lambda x: "cast(get_current_timestamp() as datetime)",
     FunctionType.DATE_TRUNCATE: lambda x: f"date_trunc('{x[1]}', {x[0]})",
-    FunctionType.DATE_ADD: lambda x: f"date_add({x[0]}, INTERVAL {x[2]} {x[1]})",
+    FunctionType.DATE_ADD: 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/parsing/common.py

@@ -100,6 +100,8 @@ def process_function_arg(
             concept.metadata.line_number = meta.line
         environment.add_concept(concept, meta=meta)
         return concept
+    elif isinstance(arg, ConceptRef):
+        return environment.concepts[arg.address]
     return arg
 
 
@@ -192,6 +194,8 @@ def concept_is_relevant(
 ) -> bool:
     if isinstance(concept, UndefinedConcept):
 
+        return False
+    if concept.datatype == DataType.UNKNOWN:
         return False
     if isinstance(concept, ConceptRef):
         if environment:

trilogy/parsing/parse_engine.py

@@ -4,7 +4,7 @@ from enum import Enum
 from os.path import dirname, join
 from pathlib import Path
 from re import IGNORECASE
-from typing import List, Optional, Tuple, Union
+from typing import Any, List, Optional, Tuple, Union
 
 from lark import Lark, ParseTree, Transformer, Tree, v_args
 from lark.exceptions import (
@@ -68,6 +68,7 @@ from trilogy.core.models.author import (
     OrderBy,
     OrderItem,
     Parenthetical,
+    RowsetItem,
     SubselectComparison,
     WhereClause,
     Window,
@@ -77,6 +78,7 @@ from trilogy.core.models.author import (
 )
 from trilogy.core.models.core import (
     DataType,
+    DataTyped,
     ListType,
     ListWrapper,
     MapType,
@@ -223,6 +225,52 @@ def unwrap_transformation(
         )
 
 
+def rehydrate_lineage(
+    lineage: Any, environment: Environment, function_factory: FunctionFactory
+) -> Any:
+    """Fix datatype propagation. This is a hack to fix the fact that we don't know the datatypes of functions until we've parsed all concepts"""
+    if isinstance(lineage, Function):
+        rehydrated = [
+            rehydrate_lineage(x, environment, function_factory)
+            for x in lineage.arguments
+        ]
+        return function_factory.create_function(
+            rehydrated,
+            operator=lineage.operator,
+        )
+    elif isinstance(lineage, Parenthetical):
+        lineage.content = rehydrate_lineage(
+            lineage.content, environment, function_factory
+        )
+        return lineage
+    elif isinstance(lineage, WindowItem):
+        lineage.content.datatype = environment.concepts[
+            lineage.content.address
+        ].datatype
+        return lineage
+    elif isinstance(lineage, AggregateWrapper):
+        lineage.function = rehydrate_lineage(
+            lineage.function, environment, function_factory
+        )
+        return lineage
+    elif isinstance(lineage, RowsetItem):
+        lineage.content.datatype = environment.concepts[
+            lineage.content.address
+        ].datatype
+        return lineage
+    else:
+        return lineage
+
+
+def rehydrate_concept_lineage(
+    concept: Concept, environment: Environment, function_factory: FunctionFactory
+) -> Concept:
+    concept.lineage = rehydrate_lineage(concept.lineage, environment, function_factory)
+    if isinstance(concept.lineage, DataTyped):
+        concept.datatype = concept.lineage.output_datatype
+    return concept
+
+
 class ParseToObjects(Transformer):
     def __init__(
         self,
@@ -275,6 +323,25 @@ class ParseToObjects(Transformer):
             v.run_second_parse_pass()
         reparsed = self.transform(self.tokens[self.token_address])
         self.environment.concepts.undefined = {}
+        passed = False
+        passes = 0
+        # output datatypes for functions may have been wrong
+        # as they were derived from not fully understood upstream types
+        # so loop through to recreate function lineage until all datatypes are known
+
+        while not passed:
+            new_passed = True
+            for x, y in self.environment.concepts.items():
+                if y.datatype == DataType.UNKNOWN and y.lineage:
+                    self.environment.concepts[x] = rehydrate_concept_lineage(
+                        y, self.environment, self.function_factory
+                    )
+                    new_passed = False
+            passes += 1
+            if passes > MAX_PARSE_DEPTH:
+                break
+            passed = new_passed
+
         return reparsed
 
     def start(self, args):
@@ -884,7 +951,8 @@ class ParseToObjects(Transformer):
         else:
             alias = self.environment.namespace
             cache_key = args[0]
-        path = args[0].split(".")
+        input_path = args[0]
+        path = input_path.split(".")
 
         if isinstance(
             self.environment.config.import_resolver, FileSystemImportResolver
@@ -904,7 +972,9 @@ class ParseToObjects(Transformer):
 
         # we don't iterate past the max parse depth
         if len(key_path) > MAX_PARSE_DEPTH:
-            return ImportStatement(alias=alias, path=Path(target))
+            return ImportStatement(
+                alias=alias, input_path=input_path, path=Path(target)
+            )
 
         if token_lookup in self.tokens:
             raw_tokens = self.tokens[token_lookup]
@@ -948,7 +1018,7 @@ class ParseToObjects(Transformer):
                 ) from e
 
         parsed_path = Path(args[0])
-        imps = ImportStatement(alias=alias, path=parsed_path)
+        imps = ImportStatement(alias=alias, input_path=input_path, path=parsed_path)
 
         self.environment.add_import(
             alias, new_env, Import(alias=alias, path=parsed_path)
@@ -1075,7 +1145,7 @@ class ParseToObjects(Transformer):
         if not select_items:
             raise ValueError("Malformed select, missing select items")
 
-        return SelectStatement.from_inputs(
+        base = SelectStatement.from_inputs(
             environment=self.environment,
             selection=select_items,
             order_by=order_by,
@@ -1084,6 +1154,7 @@ class ParseToObjects(Transformer):
             limit=limit,
             meta=Metadata(line_number=meta.line),
         )
+        return base
 
     @v_args(meta=True)
     def address(self, meta: Meta, args):
@@ -1514,6 +1585,10 @@ class ParseToObjects(Transformer):
     def fdate_add(self, meta, args):
         return self.function_factory.create_function(args, FunctionType.DATE_ADD, meta)
 
+    @v_args(meta=True)
+    def fdate_sub(self, meta, args):
+        return self.function_factory.create_function(args, FunctionType.DATE_SUB, meta)
+
     @v_args(meta=True)
     def fdate_diff(self, meta, args):
         return self.function_factory.create_function(args, FunctionType.DATE_DIFF, meta)

trilogy/parsing/trilogy.lark

@@ -293,11 +293,13 @@
     _DATE_PART.1: "date_part("i
     fdate_part: _DATE_PART expr "," DATE_PART ")"
     _DATE_ADD.1: "date_add("i
-    fdate_add: _DATE_ADD expr "," DATE_PART "," int_lit ")"
-   _DATE_DIFF.1: "date_diff("i
+    fdate_add: _DATE_ADD expr "," DATE_PART "," expr ")"
+    _DATE_SUB.1: "date_sub("i
+    fdate_sub: _DATE_SUB expr "," DATE_PART "," expr ")"
+    _DATE_DIFF.1: "date_diff("i
     fdate_diff: _DATE_DIFF expr "," expr "," DATE_PART ")"
     
-    _date_functions: fdate | fdate_add | fdate_diff | fdatetime | ftimestamp | fsecond | fminute | fhour | fday | fday_of_week | fweek | fmonth | fquarter | fyear | fdate_part | fdate_trunc
+    _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 | _generic_functions |  _date_functions