Issue #82 - Making validation errors more type-specific & adding some…

… unit tests (#122)

csvqb/csvqb/models/cube/__init__.py

@@ -2,3 +2,4 @@
 from .cube import Cube
 from .catalog import CatalogMetadataBase
 from .csvqb import *
+from .validationerrors import *

csvqb/csvqb/models/cube/csvqb/__init__.py

@@ -2,5 +2,6 @@
 from .components import *
 from .catalog import CatalogMetadata
 from ..cube import Cube
+from .validationerrors import *
 
 QbCube = Cube[CatalogMetadata]

csvqb/csvqb/models/cube/csvqb/validationerrors.py

@@ -0,0 +1,163 @@
+"""
+Qb-Cube Validation Errors
+-------------------------
+"""
+
+from dataclasses import dataclass
+from typing import Optional, Type, Union
+
+from csvqb.models.cube.csvqb.components import (
+    QbObservationValue,
+    QbMultiUnits,
+    QbDataStructureDefinition,
+)
+from csvqb.models.validationerror import SpecificValidationError
+
+ComponentTypeDescription = Union[str, Type[QbDataStructureDefinition]]
+
+
+def _get_description_for_component(t: ComponentTypeDescription) -> str:
+    if isinstance(t, str):
+        return t
+
+    return t.__name__
+
+
+@dataclass
+class OutputUriTemplateMissingError(SpecificValidationError):
+    """
+    Represents an error where the user has defined a component which cannot infer its own output_uri_template without
+    manually specifying an output_uri_template.
+    """
+
+    csv_column_name: str
+    component_type: ComponentTypeDescription
+
+    def __post_init__(self):
+        self.message = (
+            f"'{self.csv_column_name}' - a {_get_description_for_component(self.component_type)} must have an "
+            + "output_uri_template defined."
+        )
+
+
+@dataclass
+class MaxNumComponentsExceededError(SpecificValidationError):
+    """
+    Represents an error where the user can define a maximum number of components of a given type, but has defined
+    too many.
+    """
+
+    component_type: ComponentTypeDescription
+    maximum_number: int
+    actual_number: int
+    additional_explanation: Optional[str] = None
+
+    def __post_init__(self):
+        self.message = (
+            f"Found {self.actual_number} of {_get_description_for_component(self.component_type)}s. "
+            + f"Expected a maximum of {self.maximum_number}."
+        )
+        if self.additional_explanation is not None:
+            self.message += " " + self.additional_explanation
+
+
+@dataclass
+class MinNumComponentsNotSatisfiedError(SpecificValidationError):
+    """
+    Represents an error where the user must define a minimum number of components of a given type, but has not done so.
+    """
+
+    component_type: ComponentTypeDescription
+    minimum_number: int
+    actual_number: int
+    additional_explanation: Optional[str] = None
+
+    def __post_init__(self):
+        self.message = (
+            f"At least {self.minimum_number} {_get_description_for_component(self.component_type)}s must be defined."
+            + f" Found {self.actual_number}."
+        )
+        if self.additional_explanation is not None:
+            self.message += " " + self.additional_explanation
+
+
+@dataclass
+class WrongNumberComponentsError(SpecificValidationError):
+    """
+    Represents an error where the user must include a specific number of components, but has not done so.
+    """
+
+    component_type: ComponentTypeDescription
+    expected_number: int
+    actual_number: int
+    additional_explanation: Optional[str] = None
+
+    def __post_init__(self):
+        self.message = (
+            f"Found {self.actual_number} {_get_description_for_component(self.component_type)}s."
+            + f" Expected exactly {self.expected_number}."
+        )
+        if self.additional_explanation is not None:
+            self.message += " " + self.additional_explanation
+
+
+@dataclass
+class NeitherDefinedError(SpecificValidationError):
+    """
+    An error for when the user must define one of two different kinds of component, but has defined neither.
+    """
+
+    component_one: ComponentTypeDescription
+    component_two: ComponentTypeDescription
+    additional_explanation: Optional[str] = None
+
+    def __post_init__(self):
+        self.message = (
+            f"Found neither {_get_description_for_component(self.component_one)} "
+            + f"nor {_get_description_for_component(self.component_two)} defined. "
+            + "One of these must be defined."
+        )
+        if self.additional_explanation is not None:
+            self.message += " " + self.additional_explanation
+
+
+@dataclass
+class UnitsNotDefinedError(NeitherDefinedError):
+    """
+    An error for when the user has not defined any units for the dataset.
+    """
+
+    component_one: ComponentTypeDescription = f"{QbObservationValue.__name__}.unit"
+    component_two: ComponentTypeDescription = QbMultiUnits
+    additional_explanation: Optional[str] = None
+
+
+@dataclass
+class IncompatibleComponentsError(SpecificValidationError):
+    """
+    An error for when the user has defined components which are incompatible with each-other.
+    """
+
+    component_one: ComponentTypeDescription
+    component_two: ComponentTypeDescription
+    additional_explanation: Optional[str] = None
+
+    def __post_init__(self):
+        self.message = (
+            f"Both {_get_description_for_component(self.component_one)} "
+            + f"and {_get_description_for_component(self.component_two)} have been defined. "
+            + f"These components cannot be used together."
+        )
+        if self.additional_explanation is not None:
+            self.message += " " + self.additional_explanation
+
+
+@dataclass
+class BothUnitTypesDefinedError(IncompatibleComponentsError):
+    """
+    An error for when the user has both a units column as well as setting `QbObservationValue.unit`.
+    """
+
+    component_one: ComponentTypeDescription = f"{QbObservationValue.__name__}.unit"
+    component_two: ComponentTypeDescription = QbMultiUnits
+    additional_explanation: Optional[str] = None

csvqb/csvqb/models/cube/cube.py

@@ -7,9 +7,13 @@
 import pandas as pd
 
 from csvqb.models.validationerror import ValidationError
+from .validationerrors import (
+    DuplicateColumnTitleError,
+    ColumnNotFoundInDataError,
+    MissingColumnDefinitionError,
+)
 from .columns import CsvColumn
-from csvqb.models.cube.catalog import CatalogMetadataBase
-from csvqb.inputs import pandas_input_to_columnar
+from .catalog import CatalogMetadataBase
 from ..pydanticmodel import PydanticModel
 
 TMetadata = TypeVar("TMetadata", bound=CatalogMetadataBase, covariant=True)
@@ -35,9 +39,7 @@ def _validate_columns(self) -> List[ValidationError]:
         existing_col_titles: Set[str] = set()
         for col in self.columns:
             if col.csv_column_title in existing_col_titles:
-                errors.append(
-                    ValidationError(f"Duplicate column title '{col.csv_column_title}'")
-                )
+                errors.append(DuplicateColumnTitleError(col.csv_column_title))
             else:
                 existing_col_titles.add(col.csv_column_title)
 
@@ -46,22 +48,15 @@ def _validate_columns(self) -> List[ValidationError]:
                 if col.csv_column_title in self.data.columns:
                     maybe_column_data = self.data[col.csv_column_title]
                 else:
-                    errors.append(
-                        ValidationError(
-                            f"Column '{col.csv_column_title}' not found in data provided."
-                        )
-                    )
+                    errors.append(ColumnNotFoundInDataError(col.csv_column_title))
 
             errors += col.validate_data(maybe_column_data)
 
         if self.data is not None:
             defined_column_titles = [c.csv_column_title for c in self.columns]
             for column in list(self.data.columns):
+                column = str(column)
                 if column not in defined_column_titles:
-                    errors.append(
-                        ValidationError(
-                            f"Column '{column}' does not have a mapping defined."
-                        )
-                    )
+                    errors.append(MissingColumnDefinitionError(column))
 
         return errors

csvqb/csvqb/models/cube/validationerrors.py

@@ -0,0 +1,45 @@
+"""
+Cube Validation Errors
+----------------------
+"""
+from dataclasses import dataclass
+
+from csvqb.models.validationerror import SpecificValidationError
+
+
+@dataclass
+class DuplicateColumnTitleError(SpecificValidationError):
+    """
+    An error to inform the user that they have defined two instances of the same column.
+    """
+
+    csv_column_title: str
+
+    def __post_init__(self):
+        self.message = f"Duplicate column title '{self.csv_column_title}'"
+
+
+@dataclass
+class ColumnNotFoundInDataError(SpecificValidationError):
+    """
+    An error to inform the user that they have defined a column which cannot be found in the provided data.
+    """
+
+    csv_column_title: str
+
+    def __post_init__(self):
+        self.message = f"Column '{self.csv_column_title}' not found in data provided."
+
+
+@dataclass
+class MissingColumnDefinitionError(SpecificValidationError):
+    """
+    An error to inform the user that there is a column in their data that does not have a mapping specified.
+    """
+
+    csv_column_title: str
+
+    def __post_init__(self):
+        self.message = (
+            f"Column '{self.csv_column_title}' does not have a mapping defined."
+        )

csvqb/csvqb/models/rdf/__init__.py

@@ -0,0 +1,3 @@
+"""
+Contains Models for mapping data to RDF which are unique to the `csvqb` package.
+"""

csvqb/csvqb/models/validationerror.py

@@ -2,11 +2,19 @@
 ValidationError
 ---------------
 """
+from dataclasses import dataclass, field
+from abc import ABC
 
 
+@dataclass
 class ValidationError:
-    def __init__(self, message: str):
-        self.message: str = message
+    """Class representing an error validating a model."""
 
-    def __repr__(self) -> str:
-        return f"{self.__class__.__name__}({self.message})"
+    message: str
+
+
+@dataclass
+class SpecificValidationError(ValidationError, ABC):
+    """Abstract base class to represent ValidationErrors which are more specific and so can be interpreted by code."""
+
+    message: str = field(init=False)