pytrilogy 0.0.3.9__py3-none-any.whl → 0.0.3.11__py3-none-any.whl

{pytrilogy-0.0.3.9.dist-info → pytrilogy-0.0.3.11.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: pytrilogy
-Version: 0.0.3.9
+Version: 0.0.3.11
 Summary: Declarative, typed query language that compiles to SQL.
 Home-page: 
 Author: 

{pytrilogy-0.0.3.9.dist-info → pytrilogy-0.0.3.11.dist-info}/RECORD

@@ -1,31 +1,32 @@
-trilogy/__init__.py,sha256=oZcEJCenpaskC-ViU-OpC-03zsr2chE8ggGCpg2Duws,302
+trilogy/__init__.py,sha256=M6pFqPgs_BEva-ceDAVIDbK_Am1BozUgGDoN-z6dnrk,303
 trilogy/compiler.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/constants.py,sha256=qZ1d0hoKPPV2HHCoFwPYTVB7b6bXjpWvXd3lE-zEhy8,1494
 trilogy/engine.py,sha256=3etkm2RSVKO0IkgPKkrcs33X5gN_fIMyqMNfChcsR1E,1318
-trilogy/executor.py,sha256=9HhdLQoou1Cy9KSDgpdYxK6uyg-UPkgx9jXJBuK5ITc,17271
+trilogy/executor.py,sha256=CU-T7hl5hQab17KkJz9XhwlyI4-7MQL-JGdTDMVsE4E,16025
 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=6eIJewNIbrMdP6_dyLpyZ0uU62_pSj9wCP80Pkz2Bl8,1922
+trilogy/authoring/__init__.py,sha256=6KPn4uoPb_7t-i510M8UNXB7nrPpaKlxAXwuRbWBdLE,2115
 trilogy/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/core/constants.py,sha256=7XaCpZn5mQmjTobbeBn56SzPWq9eMNDfzfsRU-fP0VE,171
-trilogy/core/enums.py,sha256=aTFXL6nckmG0hpNLdCxI2kAT26cIsZPahF-pHzLQ9uc,7085
+trilogy/core/enums.py,sha256=jOgUyoLGgC5-8SLBWP0jQDMxptBJbN0J48VsCa1EMYo,7111
 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/exceptions.py,sha256=JPYyBcit3T_pRtlHdtKSeVJkIyWUTozW2aaut25A2xI,673
-trilogy/core/functions.py,sha256=rIkZGzw9hpIkXvuqQ1qPWFJO1W_NPYc6T9t0wTZ55M0,24784
+trilogy/core/functions.py,sha256=EsRYHE2kg_FckceVgYGPPs1ylrXvBjr3l1Wa6r0SvL8,25027
 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=eJOz-p20Am5IQPDeBntgIkncB1nUZGhuwcy7WBSbCmM,70240
+trilogy/core/models/author.py,sha256=CHVnj-GW_gjGDHaAUg60VGOfPSRr_5n5LbHvocOhQj8,70371
 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=yie1uuq62uOQ5fjob9NMJbdvQPrCErXUT7JTCuYRyjI,9697
+trilogy/core/models/core.py,sha256=00opIUXwgJy9OF-cwI883zQpArNeh6wkTpSqUszDU78,9909
 trilogy/core/models/datasource.py,sha256=6RjJUd2u4nYmEwFBpJlM9LbHVYDv8iHJxqiBMZqUrwI,9422
-trilogy/core/models/environment.py,sha256=qFZ0_Op6zIhKc5oVS4EVYZ67f29wJhKP_xoEMV4kkuU,25991
+trilogy/core/models/environment.py,sha256=WCVJNWUze37bxFSFWo_6Z7330OG5k0DDmb7a2Tz5gg4,26264
 trilogy/core/models/execute.py,sha256=ABylFQgtavjjCfFkEsFdUwfMB4UBQLHjdzQ9E67QlAE,33521
 trilogy/core/optimizations/__init__.py,sha256=EBanqTXEzf1ZEYjAneIWoIcxtMDite5-n2dQ5xcfUtg,356
 trilogy/core/optimizations/base_optimization.py,sha256=gzDOKImoFn36k7XBD3ysEYDnbnb6vdVIztUfFQZsGnM,513
@@ -56,29 +57,29 @@ trilogy/core/processing/node_generators/select_helpers/datasource_injection.py,s
 trilogy/core/processing/nodes/__init__.py,sha256=DqPG3Y8vl5-UTeox6hn1EE6iwPIJpsM-XeZALHSgLZQ,5058
 trilogy/core/processing/nodes/base_node.py,sha256=FHrY8GsTKPuMJklOjILbhGqCt5s1nmlj62Z-molARDA,16835
 trilogy/core/processing/nodes/filter_node.py,sha256=5VtRfKbCORx0dV-vQfgy3gOEkmmscL9f31ExvlODwvY,2461
-trilogy/core/processing/nodes/group_node.py,sha256=qwX1CaCZJzH6oYFlGRKbi-Q7KXOjfYKplz8JDU8oy5c,7998
+trilogy/core/processing/nodes/group_node.py,sha256=1caU36nHknnXS-dM6xmL9Mc7kK0wCq-IS3kL7_XYNlA,8024
 trilogy/core/processing/nodes/merge_node.py,sha256=bEz1QU2o-fl_O-VotE5dN1GmlZPClufMvUOvL2-2Uo8,15262
 trilogy/core/processing/nodes/select_node_v2.py,sha256=Xyfq8lU7rP7JTAd8VV0ATDNal64n4xIBgWQsOuMe_Ak,8824
 trilogy/core/processing/nodes/union_node.py,sha256=fDFzLAUh5876X6_NM7nkhoMvHEdGJ_LpvPokpZKOhx4,1425
 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=9wKZDwQ-BeaUCMjD9l0ffMMv8zivaYcAg12UhVFi-0Y,14248
+trilogy/core/statements/author.py,sha256=X3NdGlWTUCNdQQw9EGI1kccgsLJMazwNAIsqpUSr_ZY,14443
 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
 trilogy/dialect/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-trilogy/dialect/base.py,sha256=u00kIIl98as1QzcduiiyyoBzxRGVeBxfeO5hWlRCAJU,40222
-trilogy/dialect/bigquery.py,sha256=mKC3zoEU232h9RtIXJjqiZ72lWH8a6S28p6wAZKrAfg,2952
+trilogy/dialect/base.py,sha256=gEbbqll98nfhxs_JWK0h_9M9fh4nWnon2Ntlxp_s_lI,40287
+trilogy/dialect/bigquery.py,sha256=PkjFcNGZHYOe655PmJhb8a0afdFULuovqP0qQVO8m0I,2953
 trilogy/dialect/common.py,sha256=cbTo_vamdp8pj9spSjGSH-bSZpy4FpNJ12k5vMvyT2Y,3942
 trilogy/dialect/config.py,sha256=EGYRQIbrkeMuud5Bkds7jSD5dCJR5hEYZUYcy-lYZl4,3308
 trilogy/dialect/dataframe.py,sha256=RUbNgReEa9g3pL6H7fP9lPTrAij5pkqedpZ99D8_5AE,1522
-trilogy/dialect/duckdb.py,sha256=2tH_OetgLJoKf_f4bdeeB0ozGC8f0h_xQ271I8qD-Oo,3690
-trilogy/dialect/enums.py,sha256=1KDgds_DC31hGxZzNI_TIggxXF7m9rIjn9KLgNf5WQU,4425
+trilogy/dialect/duckdb.py,sha256=7f4qzNKXnpXA9wkU2ouarB5u2tpBC51TwLyGmN3bEG8,3693
+trilogy/dialect/enums.py,sha256=QYIcVr5RgpYMA1Wl0nWeojVVxJxy0V2_sn8uqSFNx20,4615
 trilogy/dialect/postgres.py,sha256=VH4EB4myjIeZTHeFU6vK00GxY9c53rCBjg2mLbdaCEE,3254
-trilogy/dialect/presto.py,sha256=bAxaDcLL21fivPg7hmBd3HJmd0yYJdPdwNgNA5ga7DE,3391
+trilogy/dialect/presto.py,sha256=Mw7_F8h19mWfuZHkHQJizQWbpu1lIHe6t2PA0r88gsY,3392
 trilogy/dialect/snowflake.py,sha256=wmao9p26jX5yIX5SC8sRAZTXkPGTvq6ixO693QTfhz8,2989
-trilogy/dialect/sql_server.py,sha256=IN91uEM0MpsiVAlsYC89uMQkiVn0i86B8Tst6v9uFkU,3129
+trilogy/dialect/sql_server.py,sha256=z2Vg7Qvw83rbGiEFIvHHLqVWJTWiz2xs76kpQj4HdTU,3131
 trilogy/hooks/__init__.py,sha256=T3SF3phuUDPLXKGRVE_Lf9mzuwoXWyaLolncR_1kY30,144
 trilogy/hooks/base_hook.py,sha256=I_l-NBMNC7hKTDx1JgHZPVOOCvLQ36m2oIGaR5EUMXY,1180
 trilogy/hooks/graph_hook.py,sha256=c-vC-IXoJ_jDmKQjxQyIxyXPOuUcLIURB573gCsAfzQ,2940
@@ -89,14 +90,14 @@ trilogy/parsing/common.py,sha256=IgZ2K3LzJ0usLIwxRCRmS-4luP6uwmM-f1oqGNyGbm0,213
 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=uZ6MYjg6kkTm5HFfOKLGvVvzHiGgH-vY7lV-AIlIBgY,55701
+trilogy/parsing/parse_engine.py,sha256=YjYe4cAxek3CceewGI1It-XEO_ZygCxY9IQzcJw0Y0Y,57555
 trilogy/parsing/render.py,sha256=o_XuQWhcwx1lD9eGVqkqZEwkmQK0HdmWWokGBtdeH4I,17837
-trilogy/parsing/trilogy.lark,sha256=wZpqI1louDqm-t-TpmzW749dPA9w2EIAyowyEJIeXAM,12620
+trilogy/parsing/trilogy.lark,sha256=zehaPaYKuJZQ335sgCjH8Q6u_hy5A6A02XcdwziZdWE,12817
 trilogy/scripts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 trilogy/scripts/trilogy.py,sha256=1L0XrH4mVHRt1C9T1HnaDv2_kYEfbWTb5_-cBBke79w,3774
-pytrilogy-0.0.3.9.dist-info/LICENSE.md,sha256=5ZRvtTyCCFwz1THxDTjAu3Lidds9WjPvvzgVwPSYNDo,1042
-pytrilogy-0.0.3.9.dist-info/METADATA,sha256=8Bv3VIgAPpBlaRGc1UlrWrMPLb0HYNpy-coiUXMzzK0,8983
-pytrilogy-0.0.3.9.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-pytrilogy-0.0.3.9.dist-info/entry_points.txt,sha256=0petKryjvvtEfTlbZC1AuMFumH_WQ9v8A19LvoS6G6c,54
-pytrilogy-0.0.3.9.dist-info/top_level.txt,sha256=cAy__NW_eMAa_yT9UnUNlZLFfxcg6eimUAZ184cdNiE,8
-pytrilogy-0.0.3.9.dist-info/RECORD,,
+pytrilogy-0.0.3.11.dist-info/LICENSE.md,sha256=5ZRvtTyCCFwz1THxDTjAu3Lidds9WjPvvzgVwPSYNDo,1042
+pytrilogy-0.0.3.11.dist-info/METADATA,sha256=xvuxEQpZ1puSOIBOM0_U1luF8pheKHrLQnTP3YDj-jA,8984
+pytrilogy-0.0.3.11.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+pytrilogy-0.0.3.11.dist-info/entry_points.txt,sha256=0petKryjvvtEfTlbZC1AuMFumH_WQ9v8A19LvoS6G6c,54
+pytrilogy-0.0.3.11.dist-info/top_level.txt,sha256=cAy__NW_eMAa_yT9UnUNlZLFfxcg6eimUAZ184cdNiE,8
+pytrilogy-0.0.3.11.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__ = "0.0.3.9"
+__version__ = "0.0.3.11"
 
 __all__ = ["parse", "Executor", "Dialects", "Environment", "CONFIG"]

trilogy/authoring/__init__.py

@@ -33,11 +33,20 @@ from trilogy.core.models.author import (
     WindowOrder,
     WindowType,
 )
-from trilogy.core.models.core import DataType, ListType, ListWrapper, StructType
+from trilogy.core.models.core import (
+    DataType,
+    ListType,
+    ListWrapper,
+    MapType,
+    StructType,
+)
 from trilogy.core.models.environment import Environment
 from trilogy.core.statements.author import (
     ConceptDeclarationStatement,
     ConceptTransform,
+    MultiSelectStatement,
+    PersistStatement,
+    RawSQLStatement,
     SelectItem,
     SelectStatement,
 )
@@ -60,6 +69,7 @@ __all__ = [
     "DataType",
     "StructType",
     "ListType",
+    "MapType",
     "ListWrapper",
     "FunctionType",
     "FunctionFactory",
@@ -87,4 +97,7 @@ __all__ = [
     "DEFAULT_NAMESPACE",
     "arbitrary_to_concept",
     "arg_to_datatype",
+    "MultiSelectStatement",
+    "PersistStatement",
+    "RawSQLStatement",
 ]

trilogy/core/enums.py

@@ -172,6 +172,7 @@ class FunctionType(Enum):
     UPPER = "upper"
     SUBSTRING = "substring"
     STRPOS = "strpos"
+    CONTAINS = "contains"
 
     # Dates
     DATE = "date"

trilogy/core/functions.py

@@ -271,6 +271,15 @@ FUNCTION_REGISTRY: dict[FunctionType, FunctionConfig] = {
         output_type=DataType.INTEGER,
         arg_count=2,
     ),
+    FunctionType.CONTAINS: FunctionConfig(
+        valid_inputs=[
+            {DataType.STRING},
+            {DataType.STRING},
+        ],
+        output_purpose=Purpose.PROPERTY,
+        output_type=DataType.BOOL,
+        arg_count=2,
+    ),
     FunctionType.SUBSTRING: FunctionConfig(
         valid_inputs=[{DataType.STRING}, {DataType.INTEGER}, {DataType.INTEGER}],
         output_purpose=Purpose.PROPERTY,

trilogy/core/models/author.py

@@ -55,6 +55,7 @@ from trilogy.core.models.core import (
     MapWrapper,
     NumericType,
     StructType,
+    TraitDataType,
     TupleWrapper,
     arg_to_datatype,
     is_compatible_datatype,
@@ -101,9 +102,9 @@ class HasUUID(ABC):
 
 class ConceptRef(Addressable, Namespaced, DataTyped, Mergeable, BaseModel):
     address: str
-    datatype: DataType | ListType | StructType | MapType | NumericType = (
-        DataType.UNKNOWN
-    )
+    datatype: (
+        DataType | TraitDataType | ListType | StructType | MapType | NumericType
+    ) = DataType.UNKNOWN
     metadata: Optional["Metadata"] = None
 
     @property
@@ -756,7 +757,7 @@ class Concept(Addressable, DataTyped, ConceptArgs, Mergeable, Namespaced, BaseMo
         extra="forbid",
     )
     name: str
-    datatype: DataType | ListType | StructType | MapType | NumericType
+    datatype: DataType | TraitDataType | ListType | StructType | MapType | NumericType
     purpose: Purpose
     derivation: Derivation = Derivation.ROOT
     granularity: Granularity = Granularity.MULTI_ROW
@@ -1217,9 +1218,9 @@ class UndefinedConceptFull(Concept, Mergeable, Namespaced):
     model_config = ConfigDict(arbitrary_types_allowed=True)
     name: str
     line_no: int | None = None
-    datatype: DataType | ListType | StructType | MapType | NumericType = (
-        DataType.UNKNOWN
-    )
+    datatype: (
+        DataType | TraitDataType | ListType | StructType | MapType | NumericType
+    ) = DataType.UNKNOWN
     purpose: Purpose = Purpose.UNKNOWN
 
     @property
@@ -2153,6 +2154,11 @@ class ArgBinding(BaseModel):
     default: Expr | None = None
 
 
+class CustomType(BaseModel):
+    name: str
+    type: DataType
+
+
 Expr = (
     MagicConstants
     | bool

trilogy/core/models/core.py

@@ -97,6 +97,19 @@ class DataType(Enum):
         return self
 
 
+class TraitDataType(BaseModel):
+    type: DataType
+    traits: list[str]
+
+    @property
+    def data_type(self):
+        return self.type
+
+    @property
+    def value(self):
+        return self.data_type.value
+
+
 class NumericType(BaseModel):
     precision: int = 20
     scale: int = 5

trilogy/core/models/environment.py

@@ -40,6 +40,7 @@ from trilogy.core.exceptions import (
 from trilogy.core.models.author import (
     Concept,
     ConceptRef,
+    CustomType,
     Function,
     SelectLineage,
     UndefinedConcept,
@@ -59,8 +60,23 @@ class Import:
     path: Path
 
 
+class BaseImportResolver(BaseModel):
+    pass
+
+
+class FileSystemImportResolver(BaseImportResolver):
+    pass
+
+
+class DictImportResolver(BaseImportResolver):
+    content: Dict[str, str]
+
+
 class EnvironmentOptions(BaseModel):
     allow_duplicate_declaration: bool = True
+    import_resolver: BaseImportResolver = Field(
+        default_factory=FileSystemImportResolver
+    )
 
 
 class EnvironmentConceptDict(dict):
@@ -191,14 +207,14 @@ class Environment(BaseModel):
         EnvironmentDatasourceDict, PlainValidator(validate_datasources)
     ] = Field(default_factory=EnvironmentDatasourceDict)
     functions: Dict[str, Callable[..., Any]] = Field(default_factory=dict)
-    data_types: Dict[str, DataType] = Field(default_factory=dict)
+    data_types: Dict[str, CustomType] = Field(default_factory=dict)
     named_statements: Dict[str, SelectLineage] = Field(default_factory=dict)
     imports: Dict[str, list[Import]] = Field(
         default_factory=lambda: defaultdict(list)  # type: ignore
     )
     namespace: str = DEFAULT_NAMESPACE
     working_path: str | Path = Field(default_factory=lambda: os.getcwd())
-    environment_config: EnvironmentOptions = Field(default_factory=EnvironmentOptions)
+    config: EnvironmentOptions = Field(default_factory=EnvironmentOptions)
     version: str = Field(default_factory=get_version)
     cte_name_map: Dict[str, str] = Field(default_factory=dict)
     materialized_concepts: set[str] = Field(default_factory=set)
@@ -233,7 +249,7 @@ class Environment(BaseModel):
             imports=dict(self.imports),
             namespace=self.namespace,
             working_path=self.working_path,
-            environment_config=self.environment_config,
+            environment_config=self.config,
             version=self.version,
             cte_name_map=dict(self.cte_name_map),
             materialized_concepts=set(self.materialized_concepts),
@@ -341,7 +357,7 @@ class Environment(BaseModel):
 
             return None
 
-        if existing and self.environment_config.allow_duplicate_declaration:
+        if existing and self.config.allow_duplicate_declaration:
             if existing.metadata.concept_source == ConceptSource.PERSIST_STATEMENT:
                 return handle_persist()
             return

trilogy/core/processing/nodes/group_node.py

@@ -92,7 +92,7 @@ class GroupNode(StrategyNode):
             comp_grain += source.grain
             for x in source.output_concepts:
                 concept_map[x.address] = x
-        lookups = [
+        lookups: list[BuildConcept | str] = [
             concept_map[x] if x in concept_map else x for x in comp_grain.components
         ]
         comp_grain = BuildGrain.from_concepts(lookups, environment=environment)

trilogy/core/statements/author.py

@@ -15,8 +15,11 @@ from trilogy.core.enums import (
 from trilogy.core.models.author import (
     AggregateWrapper,
     AlignClause,
+    ArgBinding,
     Concept,
     ConceptRef,
+    CustomType,
+    Expr,
     FilterItem,
     Function,
     Grain,
@@ -413,3 +416,13 @@ class ConceptDeclarationStatement(HasUUID, BaseModel):
 
 class ConceptDerivationStatement(BaseModel):
     concept: Concept
+
+
+class TypeDeclaration(BaseModel):
+    type: CustomType
+
+
+class FunctionDeclaration(BaseModel):
+    name: str
+    args: list[ArgBinding]
+    expr: Expr

trilogy/dialect/base.py

@@ -184,6 +184,7 @@ FUNCTION_MAP = {
     FunctionType.LOWER: lambda x: f"LOWER({x[0]}) ",
     FunctionType.SUBSTRING: lambda x: f"SUBSTRING({x[0]},{x[1]},{x[2]})",
     FunctionType.STRPOS: lambda x: f"STRPOS({x[0]},{x[1]})",
+    FunctionType.CONTAINS: lambda x: f"CONTAINS({x[0]},{x[1]})",
     # FunctionType.NOT_LIKE: lambda x: f" CASE WHEN {x[0]} like {x[1]} THEN 0 ELSE 1 END",
     # date types
     FunctionType.DATE_TRUNCATE: lambda x: f"date_trunc({x[0]},{x[1]})",

trilogy/dialect/bigquery.py

@@ -42,10 +42,9 @@ CREATE OR REPLACE TABLE {{ output.address.location }} AS
 {% endif %}{%- if ctes %}
 WITH {% for cte in ctes %}
 {{cte.name}} as ({{cte.statement}}){% if not loop.last %},{% endif %}{% endfor %}{% endif %}
-{%- if full_select %}
+{%- if full_select -%}
 {{full_select}}
-{% else -%}
-
+{%- else -%}
 SELECT
 {%- for select in select_columns %}
     {{ select }}{% if not loop.last %},{% endif %}{% endfor %}

trilogy/dialect/duckdb.py

@@ -57,8 +57,8 @@ CREATE OR REPLACE TABLE {{ output.address.location }} AS
 WITH {% for cte in ctes %}
 {{cte.name}} as ({{cte.statement}}){% if not loop.last %},{% endif %}{% endfor %}{% endif %}
 {%- if full_select -%}{{full_select}}
-{%- else -%}{%- if comment %}
--- {{ comment }}{% endif %}
+{%- else -%}{%- if comment -%}
+-- {{ comment }}{%- endif -%}
 SELECT
 {%- for select in select_columns %}
     {{ select }}{% if not loop.last %},{% endif %}{% endfor %}

trilogy/dialect/enums.py

@@ -41,6 +41,11 @@ class Dialects(Enum):
             return cls.DUCK_DB
         return super()._missing_(value)
 
+    def default_renderer(self, conf=None, _engine_factory: Callable = default_factory):
+        from trilogy.render import get_dialect_generator
+
+        return get_dialect_generator(self)
+
     def default_engine(self, conf=None, _engine_factory: Callable = default_factory):
         if self == Dialects.BIGQUERY:
             from google.auth import default

trilogy/dialect/presto.py

@@ -50,7 +50,7 @@ WITH {% for cte in ctes %}
 {{cte.name}} as ({{cte.statement}}){% if not loop.last %},{% endif %}{% endfor %}{% endif %}
 {%- if full_select -%}
 {{full_select}}
-{%- else %}
+{%- else -%}
 SELECT
 {%- for select in select_columns %}
     {{ select }}{% if not loop.last %},{% endif %}{% endfor %}

trilogy/dialect/sql_server.py

@@ -42,7 +42,7 @@ WITH {% for cte in ctes %}
 {{cte.name}} as ({{cte.statement}}){% if not loop.last %},{% endif %}{% endfor %}{% endif %}
 {%- if full_select -%}{{full_select}}
 {%- else -%}{%- if comment %}
--- {{ comment }}{% endif %}
+-- {{ comment }}{%- endif -%}
 SELECT
 {%- if limit is not none %}
 TOP {{ limit }}{% endif %}

trilogy/executor.py

@@ -36,6 +36,7 @@ from trilogy.dialect.enums import Dialects
 from trilogy.engine import ExecutionEngine
 from trilogy.hooks.base_hook import BaseHook
 from trilogy.parser import parse_text
+from trilogy.render import get_dialect_generator
 
 
 class ResultProtocol(Protocol):
@@ -82,40 +83,7 @@ class Executor(object):
         self.generator: BaseDialect
         self.logger = logger
         self.hooks = hooks
-        if self.dialect == Dialects.BIGQUERY:
-            from trilogy.dialect.bigquery import BigqueryDialect
-
-            self.generator = BigqueryDialect()
-        elif self.dialect == Dialects.SQL_SERVER:
-            from trilogy.dialect.sql_server import SqlServerDialect
-
-            self.generator = SqlServerDialect()
-        elif self.dialect == Dialects.DUCK_DB:
-            from trilogy.dialect.duckdb import DuckDBDialect
-
-            self.generator = DuckDBDialect()
-        elif self.dialect == Dialects.PRESTO:
-            from trilogy.dialect.presto import PrestoDialect
-
-            self.generator = PrestoDialect()
-        elif self.dialect == Dialects.TRINO:
-            from trilogy.dialect.presto import TrinoDialect
-
-            self.generator = TrinoDialect()
-        elif self.dialect == Dialects.POSTGRES:
-            from trilogy.dialect.postgres import PostgresDialect
-
-            self.generator = PostgresDialect()
-        elif self.dialect == Dialects.SNOWFLAKE:
-            from trilogy.dialect.snowflake import SnowflakeDialect
-
-            self.generator = SnowflakeDialect()
-        elif self.dialect == Dialects.DATAFRAME:
-            from trilogy.dialect.dataframe import DataframeDialect
-
-            self.generator = DataframeDialect()
-        else:
-            raise ValueError(f"Unsupported dialect {self.dialect}")
+        self.generator = get_dialect_generator(self.dialect)
         self.connection = self.engine.connect()
         # TODO: make generic
         if self.dialect == Dialects.DATAFRAME:

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 Callable, List, Optional, Tuple, Union
+from typing import List, Optional, Tuple, Union
 
 from lark import Lark, ParseTree, Transformer, Tree, v_args
 from lark.exceptions import (
@@ -57,6 +57,7 @@ from trilogy.core.models.author import (
     Concept,
     ConceptRef,
     Conditional,
+    CustomType,
     Expr,
     FilterItem,
     Function,
@@ -82,6 +83,7 @@ from trilogy.core.models.core import (
     MapWrapper,
     NumericType,
     StructType,
+    TraitDataType,
     TupleWrapper,
     arg_to_datatype,
     dict_to_map_wrapper,
@@ -95,12 +97,18 @@ from trilogy.core.models.datasource import (
     Query,
     RawColumnExpr,
 )
-from trilogy.core.models.environment import Environment, Import
+from trilogy.core.models.environment import (
+    DictImportResolver,
+    Environment,
+    FileSystemImportResolver,
+    Import,
+)
 from trilogy.core.statements.author import (
     ConceptDeclarationStatement,
     ConceptDerivationStatement,
     ConceptTransform,
     CopyStatement,
+    FunctionDeclaration,
     ImportStatement,
     Limit,
     MergeStatementV2,
@@ -111,6 +119,7 @@ from trilogy.core.statements.author import (
     SelectItem,
     SelectStatement,
     ShowStatement,
+    TypeDeclaration,
 )
 from trilogy.parsing.common import (
     align_item_to_concept,
@@ -219,7 +228,7 @@ class ParseToObjects(Transformer):
         self,
         environment: Environment,
         parse_address: str | None = None,
-        token_address: Path | None = None,
+        token_address: Path | str | None = None,
         parsed: dict[str, "ParseToObjects"] | None = None,
         tokens: dict[Path | str, ParseTree] | None = None,
         text_lookup: dict[Path | str, str] | None = None,
@@ -365,8 +374,9 @@ class ParseToObjects(Transformer):
 
     def data_type(
         self, args
-    ) -> DataType | ListType | StructType | MapType | NumericType:
+    ) -> DataType | TraitDataType | ListType | StructType | MapType | NumericType:
         resolved = args[0]
+        traits = args[2:]
         if isinstance(resolved, StructType):
             return resolved
         elif isinstance(resolved, ListType):
@@ -375,7 +385,10 @@ class ParseToObjects(Transformer):
             return resolved
         elif isinstance(resolved, MapType):
             return resolved
-        return DataType(args[0].lower())
+        base = DataType(args[0].lower())
+        if traits:
+            return TraitDataType(type=base, traits=traits)
+        return base
 
     def array_comparison(self, args) -> ComparisonOperator:
         return ComparisonOperator([x.value.lower() for x in args])
@@ -846,8 +859,22 @@ class ParseToObjects(Transformer):
         )
 
     def resolve_import_address(self, address) -> str:
-        with open(address, "r", encoding="utf-8") as f:
-            text = f.read()
+        if isinstance(
+            self.environment.config.import_resolver, FileSystemImportResolver
+        ):
+            with open(address, "r", encoding="utf-8") as f:
+                text = f.read()
+        elif isinstance(self.environment.config.import_resolver, DictImportResolver):
+            lookup = address
+            if lookup not in self.environment.config.import_resolver.content:
+                raise ImportError(
+                    f"Unable to import file {lookup}, not found in import resolver"
+                )
+            text = self.environment.config.import_resolver.content[lookup]
+        else:
+            raise ImportError(
+                f"Unable to import file {address}, resolver type {type(self.environment.config.import_resolver)} not supported"
+            )
         return text
 
     def import_statement(self, args: list[str]) -> ImportStatement:
@@ -859,10 +886,17 @@ class ParseToObjects(Transformer):
             cache_key = args[0]
         path = args[0].split(".")
 
-        target = join(self.environment.working_path, *path) + ".preql"
-
-        # tokens + text are cached by path
-        token_lookup = Path(target)
+        if isinstance(
+            self.environment.config.import_resolver, FileSystemImportResolver
+        ):
+            target = join(self.environment.working_path, *path) + ".preql"
+            # tokens + text are cached by path
+            token_lookup: Path | str = Path(target)
+        elif isinstance(self.environment.config.import_resolver, DictImportResolver):
+            target = ".".join(path)
+            token_lookup = target
+        else:
+            raise NotImplementedError
 
         # parser + env has to be cached by prior import path + current key
         key_path = self.import_keys + [cache_key]
@@ -893,6 +927,7 @@ class ParseToObjects(Transformer):
                 new_env = Environment(
                     working_path=dirname(target),
                     env_file_path=token_lookup,
+                    config=self.environment.config,
                 )
                 new_env.concepts.fail_on_missing = False
                 self.parsed[self.parse_address] = self
@@ -1087,7 +1122,7 @@ class ParseToObjects(Transformer):
         return ArgBinding(name=args[0], default=None)
 
     @v_args(meta=True)
-    def raw_function(self, meta: Meta, args) -> Callable[[list[Expr]], Expr]:
+    def raw_function(self, meta: Meta, args) -> FunctionDeclaration:
         identity = args[0]
         function_arguments: list[ArgBinding] = args[1]
         output = args[2]
@@ -1109,7 +1144,7 @@ class ParseToObjects(Transformer):
             return nout
 
         self.environment.functions[identity] = function_factory
-        return function_factory
+        return FunctionDeclaration(name=identity, args=function_arguments, expr=output)
 
     def custom_function(self, args):
         name = args[0]
@@ -1121,6 +1156,13 @@ class ParseToObjects(Transformer):
     def function(self, meta: Meta, args) -> Function:
         return args[0]
 
+    @v_args(meta=True)
+    def type_declaration(self, meta: Meta, args) -> TypeDeclaration:
+        key = args[0]
+        datatype = args[1]
+        self.environment.data_types[key] = datatype
+        return TypeDeclaration(type=CustomType(name=key, type=datatype))
+
     def int_lit(self, args):
         return int("".join(args))
 
@@ -1441,6 +1483,10 @@ class ParseToObjects(Transformer):
     def fstrpos(self, meta, args):
         return self.function_factory.create_function(args, FunctionType.STRPOS, meta)
 
+    @v_args(meta=True)
+    def fcontains(self, meta, args):
+        return self.function_factory.create_function(args, FunctionType.CONTAINS, meta)
+
     @v_args(meta=True)
     def fsubstring(self, meta, args):
         return self.function_factory.create_function(args, FunctionType.SUBSTRING, meta)

trilogy/parsing/trilogy.lark

@@ -3,6 +3,7 @@
     ?statement: concept
     | datasource
     | function
+    | type_declaration
     | multi_select_statement
     | select_statement
     | persist_statement
@@ -92,14 +93,15 @@
     function_binding_item: IDENTIFIER ("=" literal)?
     function_binding_list: (function_binding_item ",")* function_binding_item ","?
     raw_function: "def"  IDENTIFIER  "("  function_binding_list ")" "->" expr
-    
+
+    // TYPE blocks
+    type_declaration: "type" IDENTIFIER data_type 
+
     // user_id where state = Mexico
     _filter_alt: IDENTIFIER "?" conditional
     _filter_base: "filter"i IDENTIFIER where
     filter_item: _filter_base | _filter_alt
 
-
-
     // rank/lag/lead
     WINDOW_TYPE: ("row_number"i|"rank"i|"lag"i|"lead"i | "sum"i | "avg"i | "max"i | "min"i )  /[\s]+/
     
@@ -223,16 +225,18 @@
     alt_like: expr  "like"i expr
     _UPPER.1: "upper("i
     upper: _UPPER expr ")"
-   _LOWER.1: "lower("i
+    _LOWER.1: "lower("i
     lower: _LOWER expr ")"  
     _SPLIT.1: "split("i  
     fsplit: _SPLIT expr "," string_lit ")"
-   _STRPOS.1: "strpos("i
+    _STRPOS.1: "strpos("i
     fstrpos: _STRPOS expr "," expr ")"
+    _CONTAINS.1: "contains("i
+    fcontains: _CONTAINS expr "," expr ")"
     _SUBSTRING.1: "substring("i
     fsubstring: _SUBSTRING expr "," expr "," expr ")"
     
-    _string_functions: like | ilike | upper | lower | fsplit | fstrpos | fsubstring 
+    _string_functions: like | ilike | upper | lower | fsplit | fstrpos | fsubstring | fcontains
     
     // special aggregate
     _GROUP.1: "group("i
@@ -348,7 +352,7 @@
 
     map_type: "map"i "<" data_type "," data_type ">"
 
-    !data_type: "string"i | "number"i | "numeric"i | "map"i | "list"i | "array"i | "any"i | "int"i | "bigint"i | "date"i | "datetime"i | "timestamp"i | "float"i | "bool"i | numeric_type | map_type | struct_type | list_type
+    !data_type:  ("string"i | "number"i | "numeric"i | "map"i | "list"i | "array"i | "any"i | "int"i | "bigint"i | "date"i | "datetime"i | "timestamp"i | "float"i | "bool"i | numeric_type | map_type | struct_type | list_type) ("::" IDENTIFIER)?
     
     PURPOSE:  "key"i | "metric"i  | CONST
     PROPERTY: "property"i
@@ -361,7 +365,7 @@
     show_category: CONCEPTS | DATASOURCES
 
     show_statement: "show"i ( show_category | select_statement | persist_statement) _TERMINATOR
-   COMMENT: /#.*(\n|$)/ |  /\/\/.*\n/  
+    COMMENT: /#.*(\n|$)/ |  /\/\/.*\n/  
     %import common.WS
     %ignore WS
     %ignore COMMENT

trilogy/render.py

@@ -0,0 +1,38 @@
+from trilogy.dialect.enums import Dialects
+
+
+def get_dialect_generator(dialect: Dialects):
+    if dialect == Dialects.BIGQUERY:
+        from trilogy.dialect.bigquery import BigqueryDialect
+
+        return BigqueryDialect()
+    elif dialect == Dialects.SQL_SERVER:
+        from trilogy.dialect.sql_server import SqlServerDialect
+
+        return SqlServerDialect()
+    elif dialect == Dialects.DUCK_DB:
+        from trilogy.dialect.duckdb import DuckDBDialect
+
+        return DuckDBDialect()
+    elif dialect == Dialects.PRESTO:
+        from trilogy.dialect.presto import PrestoDialect
+
+        return PrestoDialect()
+    elif dialect == Dialects.TRINO:
+        from trilogy.dialect.presto import TrinoDialect
+
+        return TrinoDialect()
+    elif dialect == Dialects.POSTGRES:
+        from trilogy.dialect.postgres import PostgresDialect
+
+        return PostgresDialect()
+    elif dialect == Dialects.SNOWFLAKE:
+        from trilogy.dialect.snowflake import SnowflakeDialect
+
+        return SnowflakeDialect()
+    elif dialect == Dialects.DATAFRAME:
+        from trilogy.dialect.dataframe import DataframeDialect
+
+        return DataframeDialect()
+    else:
+        raise ValueError(f"Unsupported dialect {dialect}")