feat:rank and filter table/column neighbors

bd_api/apps/api/v1/models.py

@@ -217,6 +217,24 @@ def coverage_type(self):
 
     coverage_type.short_description = "Coverage Type"
 
+    def has_area_intersection(self, other: "Coverage"):
+        if not self.area:
+            return True
+        if not other.area:
+            return True
+        if self.area.name.startswith(other.area.name):
+            return True
+        if other.area.name.startswith(self.area.name):
+            return True
+        return False
+
+    def has_datetime_intersection(self, other: "Coverage"):
+        for dt_self in self.datetime_ranges.all():
+            for dt_other in other.datetime_ranges.all():
+                if dt_self.has_datetime_intersection(dt_other):
+                    return True
+        return False
+
     def clean(self) -> None:
         """
         Assert that only one of "table", "raw_data_source",
@@ -893,10 +911,8 @@ def clean(self) -> None:
                 "One and only one of 'table', "
                 "'raw_data_source', or 'information_request' must be set."
             )
-
         if self.entity.category.slug != "datetime":
             raise ValidationError("Entity's category is not in category.slug = `datetime`.")
-
         return super().clean()
 
 
@@ -1122,7 +1138,7 @@ def get_graphql_full_coverage(self):
     def neighbors(self):
         """Similiar tables and columns
         - Tables and columns with similar directories
-        - Tables and columns with similar coverages or tags (WIP)
+        - Tables and columns with similar coverages or tags
         """
         all_neighbors = []
         for column in self.columns.all():
@@ -1153,6 +1169,27 @@ def last_updated_at(self):
     def get_graphql_last_updated_at(self):
         return self.last_updated_at
 
+    def has_area_intersection(self, other: "Table"):
+        for cov_self in self.coverages.all():
+            for cov_other in other.coverages.all():
+                if cov_self.has_area_intersection(cov_other):
+                    return True
+        return False
+
+    def has_datetime_intersection(self, other: "Table"):
+        for cov_self in self.coverages.all():
+            for cov_other in other.coverages.all():
+                if cov_self.has_datetime_intersection(cov_other):
+                    return True
+        return False
+
+    def has_directory_intersection(self, other: "Table"):
+        for col_self in self.columns.all():
+            for col_other in other.columns.all():
+                if col_self.has_directory_intersection(col_other):
+                    return True
+        return False
+
     def clean(self):
         """
         Clean method for Table model
@@ -1275,24 +1312,6 @@ class Meta:
         verbose_name_plural = "Columns"
         ordering = ["name"]
 
-    def clean(self) -> None:
-        """Clean method for Column model"""
-        errors = {}
-        if self.observation_level and self.observation_level.table != self.table:
-            errors[
-                "observation_level"
-            ] = "Observation level is not in the same table as the column."
-
-        if self.directory_primary_key and self.directory_primary_key.table.is_directory is False:
-            errors[
-                "directory_primary_key"
-            ] = "Column indicated as a directory's primary key is not in a directory."
-
-        if errors:
-            raise ValidationError(errors)
-
-        return super().clean()
-
     @property
     def full_coverage(self) -> str:
         """
@@ -1349,7 +1368,7 @@ def get_graphql_full_coverage(self):
     def neighbors(self):
         """Similiar tables and columns
         - Columns with similar directories
-        - Columns with similar coverages or tags (WIP)
+        - Columns with similar coverages or tags
         """
         if not self.directory_primary_key:
             return []
@@ -1384,6 +1403,26 @@ def get_graphql_neighbors(self) -> list[dict]:
             )
         return get_unique_list(all_neighbors)
 
+    def has_directory_intersection(self, other: "Column"):
+        if self.directory_primary_key == other.directory_primary_key:
+            return True
+        return False
+
+    def clean(self) -> None:
+        """Clean method for Column model"""
+        errors = {}
+        if self.observation_level and self.observation_level.table != self.table:
+            errors[
+                "observation_level"
+            ] = "Observation level is not in the same table as the column."
+        if self.directory_primary_key and self.directory_primary_key.table.is_directory is False:
+            errors[
+                "directory_primary_key"
+            ] = "Column indicated as a directory's primary key is not in a directory."
+        if errors:
+            raise ValidationError(errors)
+        return super().clean()
+
 
 class ColumnOriginalName(BaseModel):
     """Model definition for ColumnOriginalName."""
@@ -1806,6 +1845,39 @@ class Meta:
         verbose_name_plural = "DateTime Ranges"
         ordering = ["id"]
 
+    @property
+    def since(self):
+        if self.start_year:
+            return datetime(
+                self.start_year,
+                self.start_month or 1,
+                self.start_day or 1,
+                self.start_hour or 0,
+                self.start_minute or 0,
+                self.start_second or 0,
+            )
+
+    @property
+    def until(self):
+        if self.end_year:
+            return datetime(
+                self.end_year,
+                self.end_month or 1,
+                self.end_day or 1,
+                self.end_hour or 0,
+                self.end_minute or 0,
+                self.end_second or 0,
+            )
+
+    def has_datetime_intersection(self, other: "DateTimeRange"):
+        if not self.since:
+            return True
+        if not other.until:
+            return True
+        if self.until >= other.since:
+            return True
+        return False
+
     def clean(self) -> None:
         """
         Assert that start_year <= end_year and start_month <= end_month