pytrilogy 0.0.3.23__tar.gz → 0.0.3.25__tar.gz

{pytrilogy-0.0.3.23/pytrilogy.egg-info → pytrilogy-0.0.3.25}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pytrilogy
-Version: 0.0.3.23
+Version: 0.0.3.25
 Summary: Declarative, typed query language that compiles to SQL.
 Home-page: 
 Author: 

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25/pytrilogy.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pytrilogy
-Version: 0.0.3.23
+Version: 0.0.3.25
 Summary: Declarative, typed query language that compiles to SQL.
 Home-page: 
 Author: 

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/tests/test_parsing.py

@@ -657,7 +657,10 @@ key x int;
     assert isinstance(env.config.import_resolver, DictImportResolver)
     env.parse(
         """
-    import test;
+import std.geography;
+import test;
+
+key fun_lat float::latitude;
                
 select x % 10 -> x_mod_10;
                

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/tests/test_typing.py

@@ -1,6 +1,25 @@
 from decimal import Decimal
+from pathlib import Path
+
+from pytest import raises
 
 from trilogy import Dialects
+from trilogy.parsing.exceptions import ParseError
+
+FILE = Path(__file__)
+
+
+def test_invalid_typing():
+    env = Dialects.DUCK_DB.default_executor()
+    with raises(ParseError):
+        env.environment.parse(
+            """
+    key customer_id int;
+    property customer_id.email string::email;
+
+
+    """
+        )
 
 
 def test_typing():
@@ -44,6 +63,58 @@ customer_id,
     assert env.environment.concepts["email"].datatype.traits == ["email"]
 
 
+def test_type_import_and_cast():
+    env = Dialects.DUCK_DB.default_executor()
+    env.environment.add_file_import(FILE.parent / "test_env_types.preql", "dtypes")
+    env.environment.parse(
+        """
+
+type year int;
+key customer_id int;
+property customer_id.signup_date date;
+
+datasource customers (
+    id:customer_id,
+    signup_date: signup_date
+)
+grain (customer_id)
+query '''
+select 1 as id, cast('2023-01-01' as date) as signup_date
+union all
+select 2 as id, cast('2023-01-02' as date) as signup_date
+''';
+
+"""
+    )
+    assert env.environment.concepts["signup_date"].keys == {"local.customer_id"}
+    assert env.environment.concepts["signup_date.year"].keys == {"local.signup_date"}
+    results = env.execute_query(
+        """SELECT
+    signup_date.year::int::year as signup_year;"""
+    )
+
+    for row in results.fetchall():
+        assert row.signup_year == 2023
+
+    assert "year" in env.environment.data_types
+
+    assert env.environment.concepts["signup_year"].datatype.traits == ["year"]
+
+    results = env.execute_query(
+        """SELECT
+signup_date.year::int::dtypes.year as signup_year_two;"""
+    )
+
+    for row in results.fetchall():
+        assert row.signup_year_two == 2023
+
+    assert "dtypes.year" in env.environment.data_types
+
+    assert env.environment.concepts["signup_year_two"].datatype.traits == [
+        "dtypes.year"
+    ]
+
+
 def test_typing_aggregate():
     env = Dialects.DUCK_DB.default_executor()
     env.environment.parse(

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/__init__.py

@@ -4,6 +4,6 @@ from trilogy.dialect.enums import Dialects
 from trilogy.executor import Executor
 from trilogy.parser import parse
 
-__version__ = "0.0.3.23"
+__version__ = "0.0.3.25"
 
 __all__ = ["parse", "Executor", "Dialects", "Environment", "CONFIG"]

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/core/environment_helpers.py

@@ -54,7 +54,7 @@ def generate_date_concepts(concept: Concept, environment: Environment):
             grain=concept.grain,
             namespace=concept.namespace,
             keys=set(
-                concept.address,
+                [concept.address],
             ),
             metadata=Metadata(
                 description=f"Auto-derived from {base_description}. {FUNCTION_DESCRIPTION_MAPS.get(ftype, ftype.value)}. ",
@@ -102,7 +102,7 @@ def generate_datetime_concepts(concept: Concept, environment: Environment):
             grain=concept.grain,
             namespace=concept.namespace,
             keys=set(
-                concept.address,
+                [concept.address],
             ),
             metadata=Metadata(
                 description=f"Auto-derived from {base_description}. {FUNCTION_DESCRIPTION_MAPS.get(ftype, ftype.value)}.",

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/core/functions.py

@@ -717,7 +717,6 @@ class FunctionFactory:
                 output_purpose = Purpose.METRIC
             else:
                 output_purpose = Purpose.PROPERTY
-
         return Function(
             operator=operator,
             arguments=full_args,

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/core/models/author.py

@@ -255,7 +255,7 @@ class Parenthetical(
         return arg_to_datatype(self.content)
 
 
-class Conditional(Mergeable, ConceptArgs, Namespaced, BaseModel):
+class Conditional(Mergeable, ConceptArgs, Namespaced, DataTyped, BaseModel):
     left: Expr
     right: Expr
     operator: BooleanOperator
@@ -357,6 +357,11 @@ class Conditional(Mergeable, ConceptArgs, Namespaced, BaseModel):
             output += self.right.existence_arguments
         return output
 
+    @property
+    def output_datatype(self):
+        # a conditional is always a boolean
+        return DataType.BOOL
+
     def decompose(self):
         chunks = []
         if self.operator == BooleanOperator.AND:
@@ -545,7 +550,7 @@ class Grain(Namespaced, BaseModel):
             return self.__add__(other)
 
 
-class Comparison(ConceptArgs, Mergeable, Namespaced, BaseModel):
+class Comparison(ConceptArgs, Mergeable, DataTyped, Namespaced, BaseModel):
     left: Union[
         int,
         str,
@@ -745,6 +750,11 @@ class Comparison(ConceptArgs, Mergeable, Namespaced, BaseModel):
             output += self.right.existence_arguments
         return output
 
+    @property
+    def output_datatype(self):
+        # a conditional is always a boolean
+        return DataType.BOOL
+
 
 class SubselectComparison(Comparison):
     def __eq__(self, other):
@@ -1035,6 +1045,7 @@ class Concept(Addressable, DataTyped, ConceptArgs, Mergeable, Namespaced, BaseMo
                     pkeys.update(parent_keys)
             raw_keys = pkeys
             # deduplicate
+
             final_grain = Grain.from_concepts(raw_keys, environment)
             keys = final_grain.components
         return new_lineage, final_grain, keys
@@ -1400,7 +1411,7 @@ def get_basic_type(
     if isinstance(type, NumericType):
         return DataType.NUMERIC
     if isinstance(type, TraitDataType):
-        return type.type
+        return get_basic_type(type.type)
     return type
 
 
@@ -2244,6 +2255,11 @@ class CustomType(BaseModel):
     name: str
     type: DataType
 
+    def with_namespace(self, namespace: str) -> "CustomType":
+        return CustomType.model_construct(
+            name=address_with_namespace(self.name, namespace), type=self.type
+        )
+
 
 Expr = (
     MagicConstants
@@ -2284,6 +2300,7 @@ FuncArgs = (
     | date
     | datetime
     | MapWrapper[Any, Any]
+    | TraitDataType
     | DataType
     | ListType
     | MapType

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/core/models/build.py

@@ -77,6 +77,7 @@ from trilogy.core.models.core import (
     MapWrapper,
     NumericType,
     StructType,
+    TraitDataType,
     TupleWrapper,
     arg_to_datatype,
 )
@@ -1066,7 +1067,9 @@ class BuildFunction(DataTyped, BuildConceptArgs, BaseModel):
     # class BuildFunction(Function):
     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[
@@ -1082,6 +1085,7 @@ class BuildFunction(DataTyped, BuildConceptArgs, BaseModel):
             date,
             datetime,
             MapWrapper[Any, Any],
+            TraitDataType,
             DataType,
             ListType,
             MapType,
@@ -1873,6 +1877,10 @@ class Factory:
         new.gen_concept_list_caches()
         return new
 
+    @build.register
+    def _(self, base: TraitDataType):
+        return base
+
     @build.register
     def _(self, base: Datasource):
         local_cache: dict[str, BuildConcept] = {}

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/core/models/core.py

@@ -105,12 +105,15 @@ class DataType(Enum):
 
 
 class TraitDataType(BaseModel):
-    type: DataType
+    type: DataType | NumericType | StructType | ListType | MapType
     traits: list[str]
 
     def __hash__(self):
         return hash(self.type)
 
+    def __str__(self) -> str:
+        return f"Trait<{self.type}, {self.traits}>"
+
     def __eq__(self, other):
         if isinstance(other, DataType):
             return self.type == other
@@ -131,6 +134,9 @@ class NumericType(BaseModel):
     precision: int = 20
     scale: int = 5
 
+    def __str__(self) -> str:
+        return f"Numeric({self.precision},{self.scale})"
+
     @property
     def data_type(self):
         return DataType.NUMERIC
@@ -353,12 +359,21 @@ def merge_datatypes(
 
 def is_compatible_datatype(left, right):
     # for unknown types, we can't make any assumptions
+    if all(
+        isinstance(x, NumericType)
+        or x in (DataType.INTEGER, DataType.FLOAT, DataType.NUMERIC)
+        for x in (left, right)
+    ):
+        return True
+    elif isinstance(left, NumericType) or isinstance(right, NumericType):
+        return False
     if right == DataType.UNKNOWN or left == DataType.UNKNOWN:
         return True
     if left == right:
         return True
     if {left, right} == {DataType.NUMERIC, DataType.FLOAT}:
         return True
+
     if {left, right} == {DataType.NUMERIC, DataType.INTEGER}:
         return True
     if {left, right} == {DataType.FLOAT, DataType.INTEGER}:
@@ -382,6 +397,8 @@ def arg_to_datatype(arg) -> CONCRETE_TYPES:
         return DataType.FLOAT
     elif isinstance(arg, NumericType):
         return arg
+    elif isinstance(arg, TraitDataType):
+        return arg
     elif isinstance(arg, ListWrapper):
         return ListType(type=arg.type)
     elif isinstance(arg, DataTyped):

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/core/models/environment.py

@@ -442,6 +442,13 @@ class Environment(BaseModel):
                 self.functions[address_with_namespace(key, alias)] = (
                     function.with_namespace(alias)
                 )
+        for key, type in source.data_types.items():
+            if same_namespace:
+                self.data_types[key] = type
+            else:
+                self.data_types[address_with_namespace(key, alias)] = (
+                    type.with_namespace(alias)
+                )
         return self
 
     def add_file_import(

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/core/processing/utility.py

@@ -38,6 +38,7 @@ from trilogy.core.models.core import (
     MapType,
     MapWrapper,
     NumericType,
+    TraitDataType,
     TupleWrapper,
 )
 from trilogy.core.models.execute import (
@@ -441,6 +442,7 @@ def is_scalar_condition(
         | BuildCaseWhen
         | BuildCaseElse
         | MagicConstants
+        | TraitDataType
         | DataType
         | MapWrapper[Any, Any]
         | ListType

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/dialect/base.py

@@ -36,6 +36,7 @@ from trilogy.core.models.core import (
     MapWrapper,
     NumericType,
     StructType,
+    TraitDataType,
     TupleWrapper,
 )
 from trilogy.core.models.datasource import Datasource, RawColumnExpr
@@ -510,6 +511,7 @@ class BaseDialect:
             date,
             datetime,
             DataType,
+            TraitDataType,
             MagicConstants,
             MapWrapper[Any, Any],
             MapType,
@@ -677,6 +679,8 @@ class BaseDialect:
             return self.FUNCTION_MAP[FunctionType.DATE_LITERAL](e)
         elif isinstance(e, datetime):
             return self.FUNCTION_MAP[FunctionType.DATETIME_LITERAL](e)
+        elif isinstance(e, TraitDataType):
+            return self.render_expr(e.type, cte=cte, cte_map=cte_map)
         else:
             raise ValueError(f"Unable to render type {type(e)} {e}")
 

{pytrilogy-0.0.3.23 → pytrilogy-0.0.3.25}/trilogy/parsing/parse_engine.py

@@ -91,6 +91,7 @@ from trilogy.core.models.core import (
     TupleWrapper,
     arg_to_datatype,
     dict_to_map_wrapper,
+    is_compatible_datatype,
     list_to_wrapper,
     tuple_to_wrapper,
 )
@@ -146,6 +147,8 @@ SELF_LABEL = "root"
 
 MAX_PARSE_DEPTH = 10
 
+STDLIB_ROOT = Path(__file__).parent.parent
+
 
 @dataclass
 class WholeGrainWrapper:
@@ -443,22 +446,36 @@ class ParseToObjects(Transformer):
     def map_type(self, args) -> MapType:
         return MapType(key_type=args[0], value_type=args[1])
 
+    @v_args(meta=True)
     def data_type(
-        self, args
+        self, meta: Meta, args
     ) -> DataType | TraitDataType | ListType | StructType | MapType | NumericType:
         resolved = args[0]
         traits = args[2:]
+        base: DataType | TraitDataType | ListType | StructType | MapType | NumericType
         if isinstance(resolved, StructType):
-            return resolved
+            base = resolved
         elif isinstance(resolved, ListType):
-            return resolved
+            base = resolved
         elif isinstance(resolved, NumericType):
-            return resolved
+            base = resolved
         elif isinstance(resolved, MapType):
-            return resolved
-        base = DataType(args[0].lower())
+            base = resolved
+        else:
+            base = DataType(args[0].lower())
         if traits:
+            for trait in traits:
+                if trait not in self.environment.data_types:
+                    raise ParseError(
+                        f"Invalid trait (type) {trait} for {base}, line {meta.line}."
+                    )
+                matched = self.environment.data_types[trait]
+                if not is_compatible_datatype(matched.type, base):
+                    raise ParseError(
+                        f"Invalid trait (type) {trait} for {base}, line {meta.line}. Trait expects type {matched.type}, has {base}"
+                    )
             return TraitDataType(type=base, traits=traits)
+
         return base
 
     def array_comparison(self, args) -> ComparisonOperator:
@@ -929,9 +946,12 @@ class ParseToObjects(Transformer):
             select=args[-1],
         )
 
-    def resolve_import_address(self, address) -> str:
-        if isinstance(
-            self.environment.config.import_resolver, FileSystemImportResolver
+    def resolve_import_address(self, address, is_stdlib: bool = False) -> str:
+        if (
+            isinstance(
+                self.environment.config.import_resolver, FileSystemImportResolver
+            )
+            or is_stdlib
         ):
             with open(address, "r", encoding="utf-8") as f:
                 text = f.read()
@@ -939,7 +959,7 @@ class ParseToObjects(Transformer):
             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"
+                    f"Unable to import file {lookup}, not found in imsport resolver"
                 )
             text = self.environment.config.import_resolver.content[lookup]
         else:
@@ -957,13 +977,17 @@ class ParseToObjects(Transformer):
             cache_key = args[0]
         input_path = args[0]
         path = input_path.split(".")
-
-        if isinstance(
+        is_stdlib = False
+        if path[0] == "std":
+            is_stdlib = True
+            target = join(STDLIB_ROOT, *path) + ".preql"
+            token_lookup: Path | str = Path(target)
+        elif 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)
+            token_lookup = Path(target)
         elif isinstance(self.environment.config.import_resolver, DictImportResolver):
             target = ".".join(path)
             token_lookup = target
@@ -984,7 +1008,7 @@ class ParseToObjects(Transformer):
             raw_tokens = self.tokens[token_lookup]
             text = self.text_lookup[token_lookup]
         else:
-            text = self.resolve_import_address(target)
+            text = self.resolve_import_address(target, is_stdlib)
             self.text_lookup[token_lookup] = text
 
             try:
@@ -1238,8 +1262,9 @@ class ParseToObjects(Transformer):
     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))
+        new = CustomType(name=key, type=datatype)
+        self.environment.data_types[key] = new
+        return TypeDeclaration(type=new)
 
     def int_lit(self, args):
         return int("".join(args))