truthound-dashboard 1.2.0__py3-none-any.whl → 1.3.0__py3-none-any.whl

truthound_dashboard/schemas/validators/table_validators.py

@@ -0,0 +1,412 @@
+"""Table validators.
+
+Validators for table-level metadata, dimensions, freshness, and structure.
+"""
+
+from .base import (
+    ParameterDefinition,
+    ParameterType,
+    ValidatorCategory,
+    ValidatorDefinition,
+)
+
+TABLE_VALIDATORS: list[ValidatorDefinition] = [
+    ValidatorDefinition(
+        name="TableNotEmpty",
+        display_name="Table Not Empty",
+        category=ValidatorCategory.TABLE,
+        description="Ensures the table contains at least one row.",
+        parameters=[],
+        tags=["table", "empty", "rows"],
+        severity_default="critical",
+    ),
+    ValidatorDefinition(
+        name="TableRowCountRange",
+        display_name="Table Row Count Range",
+        category=ValidatorCategory.TABLE,
+        description="Validates that row count falls within a range.",
+        parameters=[
+            ParameterDefinition(
+                name="min_rows",
+                label="Minimum Rows",
+                type=ParameterType.INTEGER,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="max_rows",
+                label="Maximum Rows",
+                type=ParameterType.INTEGER,
+                min_value=0,
+            ),
+        ],
+        tags=["table", "row_count", "range"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="TableRowCountExact",
+        display_name="Table Row Count Exact",
+        category=ValidatorCategory.TABLE,
+        description="Validates exact row count with tolerance.",
+        parameters=[
+            ParameterDefinition(
+                name="expected_rows",
+                label="Expected Rows",
+                type=ParameterType.INTEGER,
+                required=True,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="tolerance",
+                label="Tolerance",
+                type=ParameterType.INTEGER,
+                description="Acceptable deviation from expected",
+                default=0,
+                min_value=0,
+            ),
+        ],
+        tags=["table", "row_count", "exact"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="TableColumnCount",
+        display_name="Table Column Count",
+        category=ValidatorCategory.TABLE,
+        description="Validates the number of columns.",
+        parameters=[
+            ParameterDefinition(
+                name="expected_count",
+                label="Expected Count",
+                type=ParameterType.INTEGER,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="min_count",
+                label="Minimum Count",
+                type=ParameterType.INTEGER,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="max_count",
+                label="Maximum Count",
+                type=ParameterType.INTEGER,
+                min_value=0,
+            ),
+        ],
+        tags=["table", "column_count"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="TableRequiredColumns",
+        display_name="Table Required Columns",
+        category=ValidatorCategory.TABLE,
+        description="Validates presence of required columns.",
+        parameters=[
+            ParameterDefinition(
+                name="required_columns",
+                label="Required Columns",
+                type=ParameterType.STRING_LIST,
+                required=True,
+            ),
+        ],
+        tags=["table", "columns", "required"],
+        severity_default="critical",
+    ),
+    ValidatorDefinition(
+        name="TableForbiddenColumns",
+        display_name="Table Forbidden Columns",
+        category=ValidatorCategory.TABLE,
+        description="Ensures certain columns are absent.",
+        parameters=[
+            ParameterDefinition(
+                name="forbidden_columns",
+                label="Forbidden Columns",
+                type=ParameterType.STRING_LIST,
+                required=True,
+            ),
+        ],
+        tags=["table", "columns", "forbidden", "security"],
+        severity_default="high",
+    ),
+    ValidatorDefinition(
+        name="TableFreshness",
+        display_name="Table Freshness",
+        category=ValidatorCategory.TABLE,
+        description="Validates that the table contains recent data.",
+        parameters=[
+            ParameterDefinition(
+                name="datetime_column",
+                label="Datetime Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="max_age_hours",
+                label="Max Age (Hours)",
+                type=ParameterType.INTEGER,
+                description="Maximum age of most recent record",
+                required=True,
+                min_value=1,
+            ),
+        ],
+        tags=["table", "freshness", "recent"],
+        severity_default="high",
+    ),
+    ValidatorDefinition(
+        name="TableDataRecency",
+        display_name="Table Data Recency",
+        category=ValidatorCategory.TABLE,
+        description="Ensures data is not older than specified days.",
+        parameters=[
+            ParameterDefinition(
+                name="datetime_column",
+                label="Datetime Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="max_age_days",
+                label="Max Age (Days)",
+                type=ParameterType.INTEGER,
+                required=True,
+                min_value=1,
+            ),
+        ],
+        tags=["table", "recency", "age"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="TableUpdateFrequency",
+        display_name="Table Update Frequency",
+        category=ValidatorCategory.TABLE,
+        description="Validates expected data update frequency.",
+        parameters=[
+            ParameterDefinition(
+                name="datetime_column",
+                label="Datetime Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="expected_frequency",
+                label="Expected Frequency",
+                type=ParameterType.SELECT,
+                required=True,
+                options=[
+                    {"value": "hourly", "label": "Hourly"},
+                    {"value": "daily", "label": "Daily"},
+                    {"value": "weekly", "label": "Weekly"},
+                    {"value": "monthly", "label": "Monthly"},
+                ],
+            ),
+            ParameterDefinition(
+                name="tolerance_periods",
+                label="Tolerance Periods",
+                type=ParameterType.INTEGER,
+                description="Number of periods allowed to miss",
+                default=1,
+                min_value=0,
+            ),
+        ],
+        tags=["table", "frequency", "schedule"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="TableSchemaMatch",
+        display_name="Table Schema Match",
+        category=ValidatorCategory.TABLE,
+        description="Validates schema matches expected definition.",
+        parameters=[
+            ParameterDefinition(
+                name="expected_schema",
+                label="Expected Schema",
+                type=ParameterType.SCHEMA,
+                description="Expected schema as JSON {column: type}",
+                required=True,
+            ),
+            ParameterDefinition(
+                name="strict",
+                label="Strict Mode",
+                type=ParameterType.BOOLEAN,
+                description="Reject extra columns not in schema",
+                default=False,
+            ),
+        ],
+        tags=["table", "schema", "match"],
+        severity_default="high",
+    ),
+    ValidatorDefinition(
+        name="TableColumnTypes",
+        display_name="Table Column Types",
+        category=ValidatorCategory.TABLE,
+        description="Validates expected column data types.",
+        parameters=[
+            ParameterDefinition(
+                name="expected_types",
+                label="Expected Types",
+                type=ParameterType.SCHEMA,
+                description="JSON mapping of column names to expected types",
+                required=True,
+                placeholder='{"id": "Int64", "name": "String"}',
+            ),
+        ],
+        tags=["table", "types", "columns"],
+        severity_default="high",
+    ),
+    ValidatorDefinition(
+        name="TableMemorySize",
+        display_name="Table Memory Size",
+        category=ValidatorCategory.TABLE,
+        description="Validates estimated memory usage.",
+        parameters=[
+            ParameterDefinition(
+                name="min_size_mb",
+                label="Minimum Size (MB)",
+                type=ParameterType.FLOAT,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="max_size_mb",
+                label="Maximum Size (MB)",
+                type=ParameterType.FLOAT,
+                min_value=0,
+            ),
+        ],
+        tags=["table", "memory", "size"],
+        severity_default="low",
+    ),
+    ValidatorDefinition(
+        name="TableDimensions",
+        display_name="Table Dimensions",
+        category=ValidatorCategory.TABLE,
+        description="Validates table dimensions (rows and columns).",
+        parameters=[
+            ParameterDefinition(
+                name="min_rows",
+                label="Minimum Rows",
+                type=ParameterType.INTEGER,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="max_rows",
+                label="Maximum Rows",
+                type=ParameterType.INTEGER,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="min_cols",
+                label="Minimum Columns",
+                type=ParameterType.INTEGER,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="max_cols",
+                label="Maximum Columns",
+                type=ParameterType.INTEGER,
+                min_value=0,
+            ),
+        ],
+        tags=["table", "dimensions", "size"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="TableRowToColumnRatio",
+        display_name="Row to Column Ratio",
+        category=ValidatorCategory.TABLE,
+        description="Validates the ratio of rows to columns.",
+        parameters=[
+            ParameterDefinition(
+                name="min_ratio",
+                label="Minimum Ratio",
+                type=ParameterType.FLOAT,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="max_ratio",
+                label="Maximum Ratio",
+                type=ParameterType.FLOAT,
+                min_value=0,
+            ),
+        ],
+        tags=["table", "ratio", "dimensions"],
+        severity_default="low",
+    ),
+    ValidatorDefinition(
+        name="TablePartitionCoverage",
+        display_name="Table Partition Coverage",
+        category=ValidatorCategory.TABLE,
+        description="Validates data coverage across partitions.",
+        parameters=[
+            ParameterDefinition(
+                name="partition_column",
+                label="Partition Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="expected_partitions",
+                label="Expected Partitions",
+                type=ParameterType.STRING_LIST,
+                description="List of expected partition values",
+            ),
+            ParameterDefinition(
+                name="min_rows_per_partition",
+                label="Min Rows Per Partition",
+                type=ParameterType.INTEGER,
+                default=1,
+                min_value=0,
+            ),
+        ],
+        tags=["table", "partition", "coverage"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="TableNullRatio",
+        display_name="Table Null Ratio",
+        category=ValidatorCategory.TABLE,
+        description="Validates overall null ratio across the table.",
+        parameters=[
+            ParameterDefinition(
+                name="max_null_ratio",
+                label="Max Null Ratio",
+                type=ParameterType.FLOAT,
+                description="Maximum acceptable null ratio (0.0-1.0)",
+                default=0.1,
+                min_value=0,
+                max_value=1,
+            ),
+            ParameterDefinition(
+                name="columns",
+                label="Columns",
+                type=ParameterType.COLUMN_LIST,
+                description="Columns to check (leave empty for all)",
+            ),
+        ],
+        tags=["table", "null", "completeness"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="TableDuplicateRatio",
+        display_name="Table Duplicate Ratio",
+        category=ValidatorCategory.TABLE,
+        description="Validates overall duplicate row ratio.",
+        parameters=[
+            ParameterDefinition(
+                name="max_duplicate_ratio",
+                label="Max Duplicate Ratio",
+                type=ParameterType.FLOAT,
+                description="Maximum acceptable duplicate ratio",
+                default=0.01,
+                min_value=0,
+                max_value=1,
+            ),
+            ParameterDefinition(
+                name="columns",
+                label="Columns",
+                type=ParameterType.COLUMN_LIST,
+                description="Columns to consider for duplicates (leave empty for all)",
+            ),
+        ],
+        tags=["table", "duplicate", "uniqueness"],
+        severity_default="medium",
+    ),
+]