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

truthound_dashboard/schemas/validators/geospatial_validators.py

@@ -0,0 +1,486 @@
+"""Geospatial validators.
+
+Validators for geographic coordinate and spatial data validation.
+"""
+
+from .base import (
+    ParameterDefinition,
+    ParameterType,
+    ValidatorCategory,
+    ValidatorDefinition,
+)
+
+GEOSPATIAL_VALIDATORS: list[ValidatorDefinition] = [
+    ValidatorDefinition(
+        name="Latitude",
+        display_name="Latitude Bounds",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates latitude values are within valid bounds (-90 to 90).",
+        parameters=[
+            ParameterDefinition(
+                name="column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="min_lat",
+                label="Minimum Latitude",
+                type=ParameterType.FLOAT,
+                default=-90.0,
+                min_value=-90,
+                max_value=90,
+            ),
+            ParameterDefinition(
+                name="max_lat",
+                label="Maximum Latitude",
+                type=ParameterType.FLOAT,
+                default=90.0,
+                min_value=-90,
+                max_value=90,
+            ),
+        ],
+        tags=["geospatial", "latitude", "coordinate"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="Longitude",
+        display_name="Longitude Bounds",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates longitude values are within valid bounds (-180 to 180).",
+        parameters=[
+            ParameterDefinition(
+                name="column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="min_lon",
+                label="Minimum Longitude",
+                type=ParameterType.FLOAT,
+                default=-180.0,
+                min_value=-180,
+                max_value=180,
+            ),
+            ParameterDefinition(
+                name="max_lon",
+                label="Maximum Longitude",
+                type=ParameterType.FLOAT,
+                default=180.0,
+                min_value=-180,
+                max_value=180,
+            ),
+        ],
+        tags=["geospatial", "longitude", "coordinate"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="Coordinate",
+        display_name="Coordinate Pair",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates latitude/longitude coordinate pairs.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+        ],
+        tags=["geospatial", "coordinate", "pair"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="CoordinateNotNullIsland",
+        display_name="Not Null Island",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Detects coordinates at (0, 0) - common placeholder or error.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="tolerance",
+                label="Tolerance (degrees)",
+                type=ParameterType.FLOAT,
+                description="Tolerance around (0,0) to flag",
+                default=0.001,
+                min_value=0,
+            ),
+        ],
+        tags=["geospatial", "null_island", "quality"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="GeoDistance",
+        display_name="Distance Between Points",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates distance between two coordinate pairs.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column_a",
+                label="Latitude A",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column_a",
+                label="Longitude A",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lat_column_b",
+                label="Latitude B",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column_b",
+                label="Longitude B",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="max_distance_km",
+                label="Max Distance (km)",
+                type=ParameterType.FLOAT,
+                description="Maximum allowed distance in kilometers",
+                required=True,
+                min_value=0,
+            ),
+        ],
+        tags=["geospatial", "distance", "haversine"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="GeoDistanceFromPoint",
+        display_name="Distance From Reference Point",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates distance from a fixed reference point.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="reference_lat",
+                label="Reference Latitude",
+                type=ParameterType.FLOAT,
+                required=True,
+                min_value=-90,
+                max_value=90,
+            ),
+            ParameterDefinition(
+                name="reference_lon",
+                label="Reference Longitude",
+                type=ParameterType.FLOAT,
+                required=True,
+                min_value=-180,
+                max_value=180,
+            ),
+            ParameterDefinition(
+                name="max_distance_km",
+                label="Max Distance (km)",
+                type=ParameterType.FLOAT,
+                required=True,
+                min_value=0,
+            ),
+        ],
+        tags=["geospatial", "distance", "reference"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="GeoBoundingBox",
+        display_name="Bounding Box",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates coordinates fall within a bounding box.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="min_lat",
+                label="Min Latitude (South)",
+                type=ParameterType.FLOAT,
+                required=True,
+                min_value=-90,
+                max_value=90,
+            ),
+            ParameterDefinition(
+                name="max_lat",
+                label="Max Latitude (North)",
+                type=ParameterType.FLOAT,
+                required=True,
+                min_value=-90,
+                max_value=90,
+            ),
+            ParameterDefinition(
+                name="min_lon",
+                label="Min Longitude (West)",
+                type=ParameterType.FLOAT,
+                required=True,
+                min_value=-180,
+                max_value=180,
+            ),
+            ParameterDefinition(
+                name="max_lon",
+                label="Max Longitude (East)",
+                type=ParameterType.FLOAT,
+                required=True,
+                min_value=-180,
+                max_value=180,
+            ),
+        ],
+        tags=["geospatial", "bounding_box", "region"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="GeoCountry",
+        display_name="Country Bounds",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates coordinates fall within a country's boundaries.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="country_code",
+                label="Country Code",
+                type=ParameterType.SELECT,
+                description="ISO 3166-1 alpha-2 country code",
+                required=True,
+                options=[
+                    {"value": "US", "label": "United States"},
+                    {"value": "KR", "label": "South Korea"},
+                    {"value": "JP", "label": "Japan"},
+                    {"value": "CN", "label": "China"},
+                    {"value": "DE", "label": "Germany"},
+                    {"value": "FR", "label": "France"},
+                    {"value": "GB", "label": "United Kingdom"},
+                    {"value": "CA", "label": "Canada"},
+                    {"value": "AU", "label": "Australia"},
+                    {"value": "BR", "label": "Brazil"},
+                    {"value": "IN", "label": "India"},
+                ],
+            ),
+            ParameterDefinition(
+                name="use_approximate",
+                label="Use Approximate Bounds",
+                type=ParameterType.BOOLEAN,
+                description="Use bounding box instead of precise borders",
+                default=True,
+            ),
+        ],
+        tags=["geospatial", "country", "bounds"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="GeoPolygon",
+        display_name="Within Polygon",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates coordinates fall within a custom polygon.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="polygon",
+                label="Polygon Coordinates",
+                type=ParameterType.SCHEMA,
+                description="GeoJSON polygon coordinates",
+                required=True,
+                placeholder='[[[-122.4, 37.8], [-122.4, 37.7], [-122.3, 37.7], [-122.3, 37.8]]]',
+            ),
+        ],
+        tags=["geospatial", "polygon", "custom"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="GeoCluster",
+        display_name="Geographic Clustering",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates geographic clustering properties.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="max_cluster_radius_km",
+                label="Max Cluster Radius (km)",
+                type=ParameterType.FLOAT,
+                description="Maximum expected radius of data clusters",
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="min_points_per_cluster",
+                label="Min Points Per Cluster",
+                type=ParameterType.INTEGER,
+                default=1,
+                min_value=1,
+            ),
+        ],
+        tags=["geospatial", "clustering", "density"],
+        severity_default="low",
+    ),
+    ValidatorDefinition(
+        name="GeoSpeed",
+        display_name="Geographic Speed",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates travel speed between sequential coordinates.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="timestamp_column",
+                label="Timestamp Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="max_speed_kmh",
+                label="Max Speed (km/h)",
+                type=ParameterType.FLOAT,
+                description="Maximum realistic travel speed",
+                default=200,
+                min_value=0,
+            ),
+            ParameterDefinition(
+                name="entity_column",
+                label="Entity Column",
+                type=ParameterType.COLUMN,
+                description="Column to group by (e.g., vehicle_id)",
+            ),
+        ],
+        tags=["geospatial", "speed", "trajectory"],
+        severity_default="medium",
+    ),
+    ValidatorDefinition(
+        name="GeoAltitude",
+        display_name="Altitude Bounds",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates altitude/elevation values.",
+        parameters=[
+            ParameterDefinition(
+                name="column",
+                label="Altitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="min_altitude",
+                label="Minimum Altitude (m)",
+                type=ParameterType.FLOAT,
+                description="Minimum altitude in meters",
+                default=-500,
+            ),
+            ParameterDefinition(
+                name="max_altitude",
+                label="Maximum Altitude (m)",
+                type=ParameterType.FLOAT,
+                description="Maximum altitude in meters",
+                default=10000,
+            ),
+        ],
+        tags=["geospatial", "altitude", "elevation"],
+        severity_default="low",
+    ),
+    ValidatorDefinition(
+        name="GeoTimezone",
+        display_name="Timezone Consistency",
+        category=ValidatorCategory.GEOSPATIAL,
+        description="Validates timezone consistency with geographic location.",
+        parameters=[
+            ParameterDefinition(
+                name="lat_column",
+                label="Latitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="lon_column",
+                label="Longitude Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+            ParameterDefinition(
+                name="timezone_column",
+                label="Timezone Column",
+                type=ParameterType.COLUMN,
+                required=True,
+            ),
+        ],
+        tags=["geospatial", "timezone", "consistency"],
+        severity_default="low",
+    ),
+]