Resolve "Introduce a new filter for when rules should be run based on the source ID of a building."

Closes #21 (closed)

\approve @laurens @ds @chengzhi

rulelib/abstractrule.py

@@ -18,7 +18,6 @@
 
 
 import abc
-from typing import Union
 
 import shapely
 
@@ -29,24 +28,25 @@ class AbstractRule(abc.ABC):
     structure that is or could be necessary for all rules.
 
     Args:
-        filters_dict (dict):
-            String representing the source_id_filter for the rule. May contain more than one
-            filter input, such as the `source_id` and the `geographic_filter_boundary`.
+        rule_source_ids (list):
+            A list of all valid source IDs for the rule. This is provided by the rule's XML file
+            if relevant.
+        geographic_filter_boundary (str):
+            Boundary polygon in WKT format defining the geographic boundary of the rule.
     """
 
-    def __init__(self, filters_dict: dict = None):
-
-        rule_source_ids, geographic_filter_boundary = self.parse_filters(filters_dict)
+    def __init__(
+        self,
+        rule_source_ids: list | None = None,
+        geographic_filter_boundary: str | None = None,
+    ):
 
         self.geographic_filter_boundary_geometry = None
         if geographic_filter_boundary is not None:
-            self.geographic_filter_boundary_geometry = self.read_geometry_from_wkt(
+            self.geographic_filter_boundary_geometry = shapely.from_wkt(
                 geographic_filter_boundary
             )
 
-        if rule_source_ids is not None:
-            rule_source_ids = [int(source_id) for source_id in rule_source_ids]
-
         self.rule_source_ids = rule_source_ids
 
     @abc.abstractmethod
@@ -55,7 +55,7 @@ class AbstractRule(abc.ABC):
         The `__call__` method implements the core process of the rule.
         """
 
-    def parse_data(self, data: Union[str, bytes], source_file: str):
+    def parse_data(self, data: str | bytes, source_file: str):
         """
         The `parse_data` method can be used to parse files that are attached to the
         rule. As file types can widely vary, this method needs to be implemented for each
@@ -76,60 +76,18 @@ class AbstractRule(abc.ABC):
             + "please implement this class first."
         )
 
-    @staticmethod
-    def parse_filters(filters_dict: dict) -> tuple | None:
-        """
-        Parses the source_id_filter string input from the rule's XML file or the file attached
-        to it. Takes a single string separated by semicolons and splits it into the different
-        filter inputs it includes.
-
-        Args:
-            filters_dict (dict):
-                List with all the filter inputs.
-
-        Returns:
-            tuple | None:
-                A list of one or more strings each of which represents a different filter, or
-                None.
-        """
-
-        if bool(filters_dict):
-            if "source_id" in filters_dict:
-                source_ids = filters_dict["source_id"].split(",")
-            else:
-                source_ids = None
-            return source_ids, filters_dict["boundary"]
-
-        return None, None
-
-    @staticmethod
-    def read_geometry_from_wkt(geographic_filter_boundary):
-        """
-        Converts a WKT boundary into a Shapely polygon.
-
-        Args:
-            geographic_filter_boundary (str):
-                Boundary polygon in WKT format defining the geographic boundary of the rule.
-
-        Returns:
-            shapely.geometry.polygon.Polygon:
-                A Shapely polygon based on the WKT string geometry.
-        """
-
-        return shapely.from_wkt(geographic_filter_boundary)
-
     def filter(
         self,
-        longitude: float = None,
-        latitude: float = None,
-        source_id: int = None,
+        longitude: float | None = None,
+        latitude: float | None = None,
+        source_id: int | None = None,
         *args,
         **kwargs
     ):
         """
         Applies a spatial filter to ensure that only buildings within the provided geographic
-        polygon are processed and a source filter to ensure source specific rules only run
-        for inputs with a matching source ID.
+        polygon are processed and a source filter to ensure source-specific rules only run
+        for buildings with a matching source ID.
 
         Args:
             longitude (float):
@@ -137,24 +95,24 @@ class AbstractRule(abc.ABC):
             latitude (float):
                 Latitude of the building being processed.
             source_id (int):
-                Integer representing the data's source ID, and determines the rule's relevance.
+                Integer representing the building-source ID.
 
         Returns:
             bool:
-               True if all tests pass and False if at least one doesn't pass.
+               True if all tests pass and False if at least one does not pass.
         """
 
-        # Geographic filter, passes if the inputs coordinates lie within the boundary geometry.
+        # Geographic filter, passes if the building's coordinates lie within the boundary
+        # geometry.
         if (
             self.geographic_filter_boundary_geometry is not None
-            and not self.geographic_filter_boundary_geometry.contains(
+            and self.geographic_filter_boundary_geometry.disjoint(
                 shapely.geometry.Point(longitude, latitude)
             )
         ):
             return False
 
-        # Source ID filter, passes if the rules and the inputs source ID match.
+        # Source ID filter, passes if the rule's source ID and the building-source ID match.
         if self.rule_source_ids is not None and source_id not in self.rule_source_ids:
             return False
-
         return True