# Refactoring geometry processing

* adds GeometryProcessor
* adds GeoJsonValidator
* refactors SimpleGeomForm is_valid() logic using these new classes

konova/forms/geometry_form.py

@@ -7,9 +7,7 @@ Created on: 15.08.22
 """
 import json
 
-from django.contrib.gis import gdal
+from django.contrib.gis.geos import MultiPolygon
-from django.contrib.gis.geos import MultiPolygon, Polygon
-from django.contrib.gis.geos.prototypes.io import WKTWriter
 from django.utils.translation import gettext_lazy as _
 from django.forms import JSONField
 
@@ -18,6 +16,7 @@ from konova.models import Geometry
 from konova.settings import GEOM_MAX_VERTICES
 from konova.tasks import celery_update_parcels, celery_check_for_geometry_conflicts
 from konova.sub_settings.lanis_settings import DEFAULT_SRID_RLP
+from konova.utils.geometry.geometry_validator import GeoJsonValidator, GeometryProcessor
 from user.models import UserActionLogEntry
 
 
@@ -60,6 +59,11 @@ class SimpleGeomForm(BaseForm):
         self.initialize_form_field("output", geom)
 
     def is_valid(self):
+        """ Custom is_valid method for SimpleGeomForm
+
+        Returns:
+
+        """
         super().is_valid()
         is_valid = True
 
@@ -75,118 +79,21 @@ class SimpleGeomForm(BaseForm):
         geom = self.data.get("output", json.dumps({}))
         geom = json.loads(geom)
 
-        # Initialize features list with empty MultiPolygon, so that an empty input will result in a
+        geojson_validator = GeoJsonValidator(geom)
-        # proper empty MultiPolygon object
+        geojson_validator.validate()
-        features = []
-        features_json = geom.get("features", [])
-        accepted_ogr_types = [
-            "Polygon",
-            "Polygon25D",
-            "MultiPolygon",
-            "MultiPolygon25D",
-        ]
-        # Check validity for each feature of the geometry
-        for feature in features_json:
-            feature_geom = feature.get("geometry", feature)
-            if feature_geom is None:
-                # Fallback for rare cases where a feature does not contain any geometry
-                continue
 
-            # Try to create a geometry object from the single feature
+        if geojson_validator.errors:
-            feature_geom = json.dumps(feature_geom)
+            self.add_error("output", geojson_validator.errors)
-            g = gdal.OGRGeometry(feature_geom, srs=DEFAULT_SRID_RLP)
+            return False
 
-            geometry_has_unwanted_dimensions = g.coord_dim > 2
+        # If cleaned_data is not proper empty dict, initialize one
-            if geometry_has_unwanted_dimensions:
-                g = self.__flatten_geom_to_2D(g)
-
-            geometry_type_is_accepted = g.geom_type not in accepted_ogr_types
-            if geometry_type_is_accepted:
-                self.add_error("output", _("Only surfaces allowed. Points or lines must be buffered."))
-                is_valid &= False
-                return is_valid
-
-            is_area_valid = self.__is_area_valid(g)
-            if not is_area_valid:
-                # Geometries with an invalid size will not be saved to the db
-                # We assume these are malicious snippets which are not supposed to be in the geometry in the first place
-                self._num_geometries_ignored += 1
-                continue
-
-            # Whatever this geometry object is -> try to create a Polygon from it
-            # The resulting polygon object automatically detects whether a valid polygon has been created or not
-            g = Polygon.from_ewkt(g.ewkt)
-            is_valid &= g.valid
-            if not g.valid:
-                self.add_error("output", g.valid_reason)
-                return is_valid
-
-            # If the resulting polygon is just a single polygon, we add it to the list of properly casted features
-            if isinstance(g, Polygon):
-                features.append(g)
-            elif isinstance(g, MultiPolygon):
-                # The resulting polygon could be of type MultiPolygon (due to multiple surfaces)
-                # If so, we extract all polygons from the MultiPolygon and extend the casted features list
-                features.extend(list(g))
-
-        # Unionize all polygon features into one new MultiPolygon
-        if features:
-            form_geom = MultiPolygon(*features, srid=DEFAULT_SRID_RLP).unary_union
-        else:
-            # If no features have been processed, this indicates an empty geometry - so we store an empty geometry
-            form_geom = MultiPolygon(srid=DEFAULT_SRID_RLP)
-
-        # Make sure to convert into a MultiPolygon. Relevant if a single Polygon is provided.
-        form_geom = Geometry.cast_to_multipolygon(form_geom)
-
-        # Write unionized Multipolygon back into cleaned data
         if self.cleaned_data is None:
             self.cleaned_data = {}
-        self.cleaned_data["output"] = form_geom.ewkt
+        self.cleaned_data["output"] = geojson_validator.validated_geometry.ewkt
+        self._num_geometries_ignored = geojson_validator.num_ignored_geometries
 
         return is_valid
 
-    def __is_vertices_num_valid(self):
-        """ Checks whether the number of vertices in the geometry is not too high
-
-        Returns:
-
-        """
-        geom = self.cleaned_data.get("output")
-        g = gdal.OGRGeometry(geom, srs=DEFAULT_SRID_RLP)
-        num_vertices = g.num_coords
-
-        return num_vertices <= GEOM_MAX_VERTICES
-
-    def __is_area_valid(self, geom: gdal.OGRGeometry):
-        """ Checks whether the area is at least > 1m²
-
-        Returns:
-
-        """
-        is_area_valid = geom.area > 1  # > 1m² (SRID:25832)
-        return is_area_valid
-
-    def __simplify_geometry(self, geom, max_vert: int):
-        """ Simplifies a geometry
-
-        Geometry will be simplified until a threshold of max vertices has been reached.
-
-        Args:
-            geom (MultiPolygon): The geometry
-            max_vert (int): Threshold of maximum vertices in geometry
-
-        Returns:
-            geom (MultiPolygon): The simplified geometry
-        """
-        tolerance = 0.1
-        n = geom.num_coords
-        while(n > max_vert):
-            geom = geom.simplify(tolerance)
-            n = geom.num_coords
-            tolerance += 0.1
-        return geom
-
     def save(self, action: UserActionLogEntry):
         """ Saves the form's geometry
 
@@ -213,9 +120,9 @@ class SimpleGeomForm(BaseForm):
                 created=action,
             )
 
-        is_vertices_num_valid = self.__is_vertices_num_valid()
+        is_vertices_num_valid = GeometryProcessor.is_vertices_num_valid(geometry.geom)
         if not is_vertices_num_valid:
-            geometry.geom = self.__simplify_geometry(geometry.geom, max_vert=GEOM_MAX_VERTICES)
+            geometry.geom = GeometryProcessor.simplify_geometry(geometry.geom, max_vert=GEOM_MAX_VERTICES)
             geometry.save()
             self._geometry_simplified = True
 
@@ -240,16 +147,6 @@ class SimpleGeomForm(BaseForm):
         """
         return self._geometry_simplified
 
-    def __flatten_geom_to_2D(self, geom):
-        """
-        Enforces a given OGRGeometry from higher dimensions into 2D
-
-        """
-        wkt_w = WKTWriter(dim=2)
-        g_wkt = wkt_w.write(geom.geos).decode("utf-8")
-        geom = gdal.OGRGeometry(g_wkt)
-        return geom
-
     def _set_geojson_properties(self, geojson: dict, title: str = None):
         """ Toggles the editable property of the geojson for proper handling in map client
 

konova/utils/geometry/__init__.py

@@ -0,0 +1,5 @@
+"""
+Author: Michel Peltriaux
+Created on: 12.09.26
+
+"""

konova/utils/geometry/geometry_validator.py

@@ -0,0 +1,228 @@
+"""
+Author: Michel Peltriaux
+Created on: 12.09.26
+
+"""
+import json
+
+from django.contrib.gis import gdal
+from django.contrib.gis.gdal.srs import SpatialReference
+from django.contrib.gis.geos import Polygon, MultiPolygon
+from django.contrib.gis.geos.prototypes.io import WKTWriter
+from django.utils.translation import gettext_lazy as _
+
+from konova.settings import GEOM_MAX_VERTICES
+from konova.sub_settings.lanis_settings import DEFAULT_SRID_RLP
+
+
+class GeometryProcessor:
+    """ GeometryProcessor wraps methods for checking or processing geometry objects
+
+    """
+    @staticmethod
+    def simplify_geometry(geom: MultiPolygon, max_vert: int):
+        """ Simplifies a geometry
+
+        Geometry will be simplified until a threshold of max vertices has been reached.
+
+        Args:
+            geom (MultiPolygon): The geometry
+            max_vert (int): Threshold of maximum vertices in geometry
+
+        Returns:
+            geom (MultiPolygon): The simplified geometry
+        """
+        tolerance = 0.1
+        n = geom.num_coords
+        while(n > max_vert):
+            geom = geom.simplify(tolerance)
+            n = geom.num_coords
+            tolerance += 0.1
+        return geom
+
+    @staticmethod
+    def is_vertices_num_valid(geometry: MultiPolygon):
+        """ Checks whether the number of vertices in the geometry is not too high
+
+        Returns:
+
+        """
+        return geometry.num_coords <= GEOM_MAX_VERTICES
+
+class GeoJsonValidator:
+    """ GeoJson Validator validates geojson (e.g. from API or form input)
+
+    """
+    _errors: list
+    _input_geojson: dict
+    _validated_geometry: MultiPolygon
+    _num_ignored_geometries: int
+
+    _accepted_ogr_types = [
+        "Polygon",
+        "Polygon25D",
+        "MultiPolygon",
+        "MultiPolygon25D",
+    ]
+
+    def __init__(self, input_geojson: dict) -> None:
+        assert input_geojson is not None
+        self._input_geojson = input_geojson
+        self._num_ignored_geometries = 0
+        self._errors = list()
+
+    @property
+    def num_ignored_geometries(self):
+        """ Getter
+
+        Returns:
+
+        """
+        return self._num_ignored_geometries
+
+    @property
+    def input_geojson(self):
+        """ Getter
+
+        Returns:
+
+        """
+        return self._input_geojson
+
+    @property
+    def validated_geometry(self):
+        """ Getter
+
+        Returns:
+
+        """
+        return self._validated_geometry
+
+    @property
+    def errors(self):
+        """ Getter
+
+        Returns:
+
+        """
+        return self._errors
+
+    def __add_error(self, text: str):
+        """ Wraps pushing new error into error list
+
+        Args:
+            text (str): The error text
+
+        Returns:
+
+        """
+        self._errors.append(text)
+
+    def validate(self):
+        """ Validates input geojson
+
+        Returns:
+
+        """
+        features = self._input_geojson.get("features", [])
+        try:
+            validated_features = self.__validate_single_features(features)
+        except AssertionError as e:
+            self.__add_error(str(e))
+            return
+
+        # Unionize all polygon features into one new MultiPolygon
+        if validated_features:
+            multipolygon_geometry = MultiPolygon(*validated_features, srid=DEFAULT_SRID_RLP).unary_union
+        else:
+            # If no features have been processed, this indicates an empty geometry - so we store an empty geometry
+            multipolygon_geometry = MultiPolygon(srid=DEFAULT_SRID_RLP)
+
+        # Make sure to convert into a MultiPolygon. Relevant if a single Polygon is provided.
+        multipolygon_geometry = self.cast_to_multipolygon(multipolygon_geometry)
+        self._validated_geometry = multipolygon_geometry
+
+    def __validate_single_features(self, features: list) -> list:
+        validated_features = []
+
+        srs = SpatialReference(DEFAULT_SRID_RLP)
+        # Check validity for each feature of the geometry
+        for feature in features:
+            feature_geom = feature.get("geometry", feature)
+            if feature_geom is None:
+                # Fallback for rare cases where a feature does not contain any geometry
+                continue
+
+            # Try to create a geometry object from the single feature
+            feature_geom = json.dumps(feature_geom)
+            g = gdal.OGRGeometry(feature_geom, srs=srs)
+
+            geometry_has_unwanted_dimensions = g.coord_dim > 2
+            if geometry_has_unwanted_dimensions:
+                g = self.__flatten_geom_to_2D(g)
+
+            geometry_type_is_accepted = g.geom_type not in self._accepted_ogr_types
+            if geometry_type_is_accepted:
+                raise AssertionError(_("Only surfaces allowed. Points or lines must be buffered."))
+
+            is_area_valid = self.__is_area_valid(g)
+            if not is_area_valid:
+                # Geometries with an invalid size will not be saved to the db
+                # We assume these are malicious snippets which are not supposed to be in the geometry in the first place
+                self._num_ignored_geometries += 1
+                continue
+
+            # Whatever this geometry object is -> try to create a Polygon from it
+            # The resulting polygon object automatically detects whether a valid polygon has been created or not
+            g = Polygon.from_ewkt(g.ewkt)
+            if not g.valid:
+                raise AssertionError(g.valid_reason)
+
+            # If the resulting polygon is just a single polygon, we add it to the list of properly created features
+            if isinstance(g, Polygon):
+                validated_features.append(g)
+            elif isinstance(g, MultiPolygon):
+                # The resulting polygon could be of type MultiPolygon (due to multiple surfaces)
+                # If so, we extract each single polygons from the MultiPolygon and add them to the created features list
+                validated_features.extend(list(g))
+
+        return validated_features
+
+
+    def __flatten_geom_to_2D(self, geom: gdal.OGRGeometry) -> gdal.OGRGeometry:
+        """
+        Enforces a given OGRGeometry from higher dimensions into 2D
+
+        """
+        wkt_w = WKTWriter(dim=2)
+        g_wkt = wkt_w.write(geom.geos).decode("utf-8")
+        geom = gdal.OGRGeometry(g_wkt)
+        return geom
+
+    def __is_area_valid(self, geom: gdal.OGRGeometry):
+        """ Checks whether the area is at least > 1m²
+
+        Returns:
+
+        """
+        is_area_valid = geom.area > 1  # > 1m² (SRID:25832)
+        return is_area_valid
+
+    @staticmethod
+    def cast_to_multipolygon(input_geom):
+        """ If input_geom is not a MultiPolygon, cast to MultiPolygon
+
+        Args:
+            input_geom ():
+
+        Returns:
+            output_geom
+        """
+        output_geom = input_geom
+        if not isinstance(input_geom, MultiPolygon):
+            try:
+                output_geom = MultiPolygon(input_geom, srid=DEFAULT_SRID_RLP)
+            except TypeError as e:
+                raise AssertionError(f"Only (Multi)Polygon allowed! Could not convert {input_geom.geom_type} to MultiPolygon")
+        return output_geom
+