vtlengine 1.1rc2__py3-none-any.whl → 1.2.0__py3-none-any.whl

vtlengine/API/_InternalApi.py

@@ -1,18 +1,37 @@
 import json
 import os
 from pathlib import Path
-from typing import Any, Dict, List, Optional, Union
+from typing import Any, Dict, List, Literal, Optional, Tuple, Union
 
 import jsonschema
 import pandas as pd
+from pysdmx.model.dataflow import Component as SDMXComponent
+from pysdmx.model.dataflow import DataStructureDefinition, Schema
+from pysdmx.model.dataflow import Role as SDMX_Role
+from pysdmx.model.vtl import (
+    Ruleset,
+    RulesetScheme,
+    Transformation,
+    TransformationScheme,
+    UserDefinedOperator,
+    UserDefinedOperatorScheme,
+)
 
+from vtlengine import AST as AST
 from vtlengine.__extras_check import __check_s3_extra
-from vtlengine.AST import PersistentAssignment, Start
+from vtlengine.AST import Assignment, DPRuleset, HRuleset, Operator, PersistentAssignment, Start
+from vtlengine.AST.ASTString import ASTString
 from vtlengine.DataTypes import SCALAR_TYPES
-from vtlengine.Exceptions import InputValidationException, check_key
+from vtlengine.Exceptions import (
+    InputValidationException,
+    SemanticError,
+    check_key,
+)
 from vtlengine.files.parser import _fill_dataset_empty_data, _validate_pandas
 from vtlengine.Model import (
-    Component,
+    Component as VTL_Component,
+)
+from vtlengine.Model import (
     Dataset,
     ExternalRoutine,
     Role,
@@ -20,18 +39,23 @@ from vtlengine.Model import (
     Scalar,
     ValueDomain,
 )
+from vtlengine.Utils import VTL_DTYPES_MAPPING, VTL_ROLE_MAPPING
 
 base_path = Path(__file__).parent
 schema_path = base_path / "data" / "schema"
+sdmx_csv_path = base_path / "data" / "sdmx_csv"
 with open(schema_path / "json_schema_2.1.json", "r") as file:
     schema = json.load(file)
 
 
-def _load_dataset_from_structure(structures: Dict[str, Any]) -> Dict[str, Any]:
+def _load_dataset_from_structure(
+    structures: Dict[str, Any],
+) -> Tuple[Dict[str, Any], Dict[str, Any]]:
     """
     Loads a dataset with the structure given.
     """
     datasets = {}
+    scalars = {}
 
     if "datasets" in structures:
         for dataset_json in structures["datasets"]:
@@ -66,7 +90,7 @@ def _load_dataset_from_structure(structures: Dict[str, Any]) -> Dict[str, Any]:
                         else:
                             component["nullable"] = False
 
-                    components[component["name"]] = Component(
+                    components[component["name"]] = VTL_Component(
                         name=component["name"],
                         data_type=SCALAR_TYPES[component["data_type"]],
                         role=Role(component["role"]),
@@ -77,7 +101,7 @@ def _load_dataset_from_structure(structures: Dict[str, Any]) -> Dict[str, Any]:
                 for component in dataset_json["DataStructure"]:
                     check_key("data_type", SCALAR_TYPES.keys(), component["type"])
                     check_key("role", Role_keys, component["role"])
-                    components[component["name"]] = Component(
+                    components[component["name"]] = VTL_Component(
                         name=component["name"],
                         data_type=SCALAR_TYPES[component["type"]],
                         role=Role(component["role"]),
@@ -93,8 +117,8 @@ def _load_dataset_from_structure(structures: Dict[str, Any]) -> Dict[str, Any]:
                 data_type=SCALAR_TYPES[scalar_json["type"]],
                 value=None,
             )
-            datasets[scalar_name] = scalar  # type: ignore[assignment]
-    return datasets
+            scalars[scalar_name] = scalar
+    return datasets, scalars
 
 
 def _load_single_datapoint(datapoint: Union[str, Path]) -> Dict[str, Any]:
@@ -142,7 +166,9 @@ def _load_datapoints_path(
     return _load_single_datapoint(datapoints)
 
 
-def _load_datastructure_single(data_structure: Union[Dict[str, Any], Path]) -> Dict[str, Dataset]:
+def _load_datastructure_single(
+    data_structure: Union[Dict[str, Any], Path],
+) -> Tuple[Dict[str, Dataset], Dict[str, Scalar]]:
     """
     Loads a single data structure.
     """
@@ -153,13 +179,15 @@ def _load_datastructure_single(data_structure: Union[Dict[str, Any], Path]) -> D
     if not data_structure.exists():
         raise Exception("Invalid datastructure. Input does not exist")
     if data_structure.is_dir():
-        datasets: Dict[str, Any] = {}
+        datasets: Dict[str, Dataset] = {}
+        scalars: Dict[str, Scalar] = {}
         for f in data_structure.iterdir():
             if f.suffix != ".json":
                 continue
-            dataset = _load_datastructure_single(f)
-            datasets = {**datasets, **dataset}
-        return datasets
+            ds, sc = _load_datastructure_single(f)
+            datasets = {**datasets, **ds}
+            scalars = {**scalars, **sc}
+        return datasets, scalars
     else:
         if data_structure.suffix != ".json":
             raise Exception("Invalid datastructure. Must have .json extension")
@@ -169,8 +197,8 @@ def _load_datastructure_single(data_structure: Union[Dict[str, Any], Path]) -> D
 
 
 def load_datasets(
-    data_structure: Union[Dict[str, Any], Path, List[Union[Dict[str, Any], Path]]],
-) -> Dict[str, Dataset]:
+    data_structure: Union[Dict[str, Any], Path, List[Dict[str, Any]], List[Path]],
+) -> Tuple[Dict[str, Dataset], Dict[str, Scalar]]:
     """
     Loads multiple datasets.
 
@@ -188,21 +216,42 @@ def load_datasets(
     if isinstance(data_structure, dict):
         return _load_datastructure_single(data_structure)
     if isinstance(data_structure, list):
-        ds_structures: Dict[str, Any] = {}
+        ds_structures: Dict[str, Dataset] = {}
+        scalar_structures: Dict[str, Scalar] = {}
         for x in data_structure:
-            result = _load_datastructure_single(x)
-            ds_structures = {**ds_structures, **result}  # Overwrite ds_structures dict.
-        return ds_structures
+            ds, sc = _load_datastructure_single(x)
+            ds_structures = {**ds_structures, **ds}  # Overwrite ds_structures dict.
+            scalar_structures = {**scalar_structures, **sc}  # Overwrite scalar_structures dict.
+        return ds_structures, scalar_structures
     return _load_datastructure_single(data_structure)
 
 
-def load_datasets_with_data(data_structures: Any, datapoints: Optional[Any] = None) -> Any:
+def _handle_scalars_values(
+    scalars: Dict[str, Scalar],
+    scalar_values: Optional[Dict[str, Optional[Union[int, str, bool, float]]]] = None,
+) -> None:
+    if scalar_values is None:
+        return
+    # Handling scalar values with the scalar dict
+    for name, value in scalar_values.items():
+        if name not in scalars:
+            raise Exception(f"Not found scalar {name} in datastructures")
+        # Casting value to scalar data type
+        scalars[name].value = scalars[name].data_type.cast(value)
+
+
+def load_datasets_with_data(
+    data_structures: Any,
+    datapoints: Optional[Any] = None,
+    scalar_values: Optional[Dict[str, Optional[Union[int, str, bool, float]]]] = None,
+) -> Any:
     """
     Loads the dataset structures and fills them with the data contained in the datapoints.
 
     Args:
         data_structures: Dict, Path or a List of dicts or Paths.
         datapoints: Dict, Path or a List of Paths.
+        scalar_values: Dict with the scalar values.
 
     Returns:
         A dict with the structure and a pandas dataframe with the data.
@@ -210,17 +259,18 @@ def load_datasets_with_data(data_structures: Any, datapoints: Optional[Any] = No
     Raises:
         Exception: If the Path is wrong or the file is invalid.
     """
-    datasets = load_datasets(data_structures)
+    datasets, scalars = load_datasets(data_structures)
     if datapoints is None:
         for dataset in datasets.values():
             if isinstance(dataset, Dataset):
                 _fill_dataset_empty_data(dataset)
-        return datasets, None
+        _handle_scalars_values(scalars, scalar_values)
+        return datasets, scalars, None
     if isinstance(datapoints, dict):
         # Handling dictionary of Pandas Dataframes
         for dataset_name, data in datapoints.items():
             if dataset_name not in datasets:
-                raise Exception(f"Not found dataset {dataset_name}")
+                raise Exception(f"Not found dataset {dataset_name} in datastructures.")
             datasets[dataset_name].data = _validate_pandas(
                 datasets[dataset_name].components, data, dataset_name
             )
@@ -229,14 +279,17 @@ def load_datasets_with_data(data_structures: Any, datapoints: Optional[Any] = No
                 datasets[dataset_name].data = pd.DataFrame(
                     columns=list(datasets[dataset_name].components.keys())
                 )
-        return datasets, None
+        _handle_scalars_values(scalars, scalar_values)
+        return datasets, scalars, None
     # Handling dictionary of paths
     dict_datapoints = _load_datapoints_path(datapoints)
     for dataset_name, _ in dict_datapoints.items():
         if dataset_name not in datasets:
-            raise Exception(f"Not found dataset {dataset_name}")
+            raise Exception(f"Not found dataset {dataset_name} in datastructures.")
 
-    return datasets, dict_datapoints
+    _handle_scalars_values(scalars, scalar_values)
+
+    return datasets, scalars, dict_datapoints
 
 
 def load_vtl(input: Union[str, Path]) -> str:
@@ -345,8 +398,8 @@ def load_external_routines(input: Union[Dict[str, Any], Path, str]) -> Any:
 
 
 def _return_only_persistent_datasets(
-    datasets: Dict[str, Dataset], ast: Start
-) -> Dict[str, Dataset]:
+    datasets: Dict[str, Union[Dataset, Scalar]], ast: Start
+) -> Dict[str, Union[Dataset, Scalar]]:
     """
     Returns only the datasets with a persistent assignment.
     """
@@ -393,3 +446,209 @@ def _check_output_folder(output_folder: Union[str, Path]) -> None:
         if output_folder.suffix != "":
             raise ValueError("Output folder must be a Path or S3 URI to a directory")
         os.mkdir(output_folder)
+
+
+def to_vtl_json(dsd: Union[DataStructureDefinition, Schema], dataset_name: str) -> Dict[str, Any]:
+    """
+    Converts a pysdmx `DataStructureDefinition` or `Schema` into a VTL-compatible JSON
+    representation.
+
+    This function extracts and transforms the components (dimensions, measures, and attributes)
+    from the given SDMX data structure and maps them into a dictionary format that conforms
+    to the expected VTL data structure json schema.
+
+    Args:
+        dsd: An instance of `DataStructureDefinition` or `Schema` from the `pysdmx` model.
+        dataset_name: The name of the resulting VTL dataset.
+
+    Returns:
+            A dictionary representing the dataset in VTL format, with keys for dataset name and its
+            components, including their name, role, data type, and nullability.
+    """
+    components = []
+    NAME = "name"
+    ROLE = "role"
+    TYPE = "type"
+    NULLABLE = "nullable"
+
+    _components: List[SDMXComponent] = []
+    _components.extend(dsd.components.dimensions)
+    _components.extend(dsd.components.measures)
+    _components.extend(dsd.components.attributes)
+
+    for c in _components:
+        _type = VTL_DTYPES_MAPPING[c.dtype]
+        _nullability = c.role != SDMX_Role.DIMENSION
+        _role = VTL_ROLE_MAPPING[c.role]
+
+        component = {
+            NAME: c.id,
+            ROLE: _role,
+            TYPE: _type,
+            NULLABLE: _nullability,
+        }
+
+        components.append(component)
+
+    result = {"datasets": [{"name": dataset_name, "DataStructure": components}]}
+
+    return result
+
+
+def __generate_transformation(
+    child: Union[Assignment, PersistentAssignment], is_persistent: bool, count: int
+) -> Transformation:
+    expression = ASTString().render(ast=child.right)
+    result = child.left.value  # type: ignore[attr-defined]
+    return Transformation(
+        id=f"T{count}",
+        expression=expression,
+        is_persistent=is_persistent,
+        result=result,
+        name=f"Transformation {result}",
+    )
+
+
+def __generate_udo(child: Operator, count: int) -> UserDefinedOperator:
+    operator_definition = ASTString().render(ast=child)
+    return UserDefinedOperator(
+        id=f"UDO{count}",
+        operator_definition=operator_definition,
+        name=f"UserDefinedOperator {child.op}",
+    )
+
+
+def __generate_ruleset(child: Union[DPRuleset, HRuleset], count: int) -> Ruleset:
+    ruleset_definition = ASTString().render(ast=child)
+    ruleset_type: Literal["datapoint", "hierarchical"] = (
+        "datapoint" if isinstance(child, DPRuleset) else "hierarchical"
+    )
+    ruleset_scope: Literal["variable", "valuedomain"] = (
+        "variable" if child.signature_type == "variable" else "valuedomain"
+    )
+    return Ruleset(
+        id=f"R{count}",
+        ruleset_definition=ruleset_definition,
+        ruleset_type=ruleset_type,
+        ruleset_scope=ruleset_scope,
+        name=f"{ruleset_type.capitalize()} ruleset {child.name}",
+    )
+
+
+def ast_to_sdmx(ast: AST.Start, agency_id: str, id: str, version: str) -> TransformationScheme:
+    """
+    Converts a vtl AST into an SDMX compatible `TransformationScheme` object, following
+    the pysdmx model.
+
+    This function iterates over the child nodes of the given AST and categorizes each into one of
+    the following types:
+    - `PersistentAssignment`: Represents a persistent transformation. These are added to the
+    transformation list with a persistence flag.
+    - `Assignment`: Represents a temporary (non-persistent) transformation. These are added to the
+    transformation list without the persistence flag
+    - `DPRuleset` or `HRuleset`: Represent validation rule sets.
+    These are collected and wrapped into a `RulesetScheme` object.
+    - `Operator`: Defines user-defined operators. These are collected
+    into a `UserDefinedOperatorScheme` object.
+
+    After parsing all AST elements:
+    - If any rulesets were found, a `RulesetScheme` is created and added to the references.
+    - If any user-defined operators were found, a `UserDefinedOperatorScheme` is created and added
+    to the references.
+    - A `TransformationScheme` object is constructed with all collected transformations and any
+    additional references.
+
+    Args:
+        ast: The root node of the vtl ast representing the set of
+        vtl expressions.
+        agency_id: The identifier of the agency defining the SDMX structure as a string.
+        id: The identifier of the transformation scheme as a string.
+        version: The version of the transformation scheme given as a string.
+
+    Returns:
+        TransformationScheme: A fully constructed transformation scheme that includes
+        transformations, and optionally rule sets and user-defined operator schemes,
+        suitable for SDMX.
+
+    """
+    list_transformation = []
+    list_udos = []
+    list_rulesets = []
+    count_transformation = 0
+    count_udo = 0
+    count_ruleset = 0
+
+    for child in ast.children:
+        if isinstance(child, PersistentAssignment):
+            count_transformation += 1
+            list_transformation.append(
+                __generate_transformation(
+                    child=child, is_persistent=True, count=count_transformation
+                )
+            )
+        elif isinstance(child, Assignment):
+            count_transformation += 1
+            list_transformation.append(
+                __generate_transformation(
+                    child=child, is_persistent=False, count=count_transformation
+                )
+            )
+        elif isinstance(child, (DPRuleset, HRuleset)):
+            count_ruleset += 1
+            list_rulesets.append(__generate_ruleset(child=child, count=count_ruleset))
+        elif isinstance(child, Operator):
+            count_udo += 1
+            list_udos.append(__generate_udo(child=child, count=count_udo))
+
+    references: Any = {}
+    if list_rulesets:
+        references["ruleset_schemes"] = [
+            RulesetScheme(
+                items=list_rulesets,
+                agency=agency_id,
+                id="RS1",
+                vtl_version="2.1",
+                version=version,
+                name=f"RulesetScheme {id}-RS",
+            )
+        ]
+    if list_udos:
+        references["user_defined_operator_schemes"] = [
+            UserDefinedOperatorScheme(
+                items=list_udos,
+                agency=agency_id,
+                id="UDS1",
+                vtl_version="2.1",
+                version=version,
+                name=f"UserDefinedOperatorScheme {id}-UDS",
+            )
+        ]
+
+    transformation_scheme = TransformationScheme(
+        items=list_transformation,
+        agency=agency_id,
+        id="TS1",
+        vtl_version="2.1",
+        version=version,
+        name=f"TransformationScheme {id}",
+        **references,
+    )
+
+    return transformation_scheme
+
+
+def _check_script(script: Union[str, TransformationScheme, Path]) -> str:
+    """
+    Check if the TransformationScheme object is valid to generate a vtl script.
+    """
+    if not isinstance(script, (str, TransformationScheme, Path)):
+        raise SemanticError("0-1-1-1", format_=type(script).__name__)
+    if isinstance(script, TransformationScheme):
+        from pysdmx.toolkit.vtl import (
+            generate_vtl_script,
+        )
+
+        vtl_script = generate_vtl_script(script, model_validation=True)
+        return vtl_script
+    else:
+        return str(script)