Merge pull request #123 from stac-utils/bbox=geometry

- Added check to verify that bbox matches item's polygon geometry

Author: jonhealy1

CHANGELOG.md

@@ -9,6 +9,7 @@ The format is (loosely) based on [Keep a Changelog](http://keepachangelog.com/)
 ### Added
 
 - Added sponsors and supporters section with logos ([#122](https://github.com/stac-utils/stac-check/pull/122))
+- Added check to verify that bbox matches item's polygon geometry ([#123](https://github.com/stac-utils/stac-check/pull/123))
 - Added configuration documentation to README ([#124](https://github.com/stac-utils/stac-check/pull/124))
 
 ### Updated

stac_check/lint.py

@@ -450,10 +450,68 @@ def check_geometry_null(self) -> bool:
             bool: A boolean indicating whether the geometry property is null (True) or not (False).
         """
         if "geometry" in self.data:
-            return self.data["geometry"] is None
+            return self.data.get("geometry") is None
         else:
             return False
 
+    def check_bbox_matches_geometry(self) -> bool:
+        """Checks if the bbox of a STAC item matches its geometry.
+
+        This function verifies that the bounding box (bbox) accurately represents
+        the minimum bounding rectangle of the item's geometry. It only applies to
+        items with non-null geometry of type Polygon or MultiPolygon.
+
+        Returns:
+            bool: True if the bbox matches the geometry or if the check is not applicable
+                 (e.g., null geometry or non-polygon type). False if there's a mismatch.
+        """
+        # Skip check if geometry is null or bbox is not present
+        if (
+            "geometry" not in self.data
+            or self.data.get("geometry") is None
+            or "bbox" not in self.data
+            or self.data.get("bbox") is None
+        ):
+            return True
+
+        geometry = self.data.get("geometry")
+        bbox = self.data.get("bbox")
+
+        # Only process Polygon and MultiPolygon geometries
+        geom_type = geometry.get("type")
+        if geom_type not in ["Polygon", "MultiPolygon"]:
+            return True
+
+        # Extract coordinates based on geometry type
+        coordinates = []
+        if geom_type == "Polygon":
+            # For Polygon, use the exterior ring (first element)
+            if len(geometry.get("coordinates", [])) > 0:
+                coordinates = geometry.get("coordinates")[0]
+        elif geom_type == "MultiPolygon":
+            # For MultiPolygon, collect all coordinates from all polygons
+            for polygon in geometry.get("coordinates", []):
+                if len(polygon) > 0:
+                    coordinates.extend(polygon[0])
+
+        # If no valid coordinates, skip check
+        if not coordinates:
+            return True
+
+        # Calculate min/max from coordinates
+        lons = [coord[0] for coord in coordinates]
+        lats = [coord[1] for coord in coordinates]
+
+        calc_bbox = [min(lons), min(lats), max(lons), max(lats)]
+
+        # Allow for small floating point differences (epsilon)
+        epsilon = 1e-8
+        for i in range(4):
+            if abs(bbox[i] - calc_bbox[i]) > epsilon:
+                return False
+
+        return True
+
     def check_searchable_identifiers(self) -> bool:
         """Checks if the identifiers of a STAC item are searchable, i.e.,
         they only contain lowercase letters, numbers, hyphens, and underscores.
@@ -616,6 +674,14 @@ def create_best_practices_dict(self) -> Dict:
             msg_1 = "All items should have a geometry field. STAC is not meant for non-spatial data"
             best_practices_dict["null_geometry"] = [msg_1]
 
+        # best practices - check if bbox matches geometry
+        if (
+            not self.check_bbox_matches_geometry()
+            and config.get("check_bbox_geometry_match", True) == True
+        ):
+            msg_1 = "The bbox field does not match the bounds of the geometry. The bbox should be the minimum bounding rectangle of the geometry."
+            best_practices_dict["bbox_geometry_mismatch"] = [msg_1]
+
         # check to see if there are too many links
         if (
             self.check_bloated_links(max_links=max_links)

stac_check/stac-check.config.yml

@@ -15,6 +15,8 @@ linting:
   check_unlocated: true
   # best practices - recommend items have a geometry
   check_geometry: true
+  # best practices - check if bbox matches the bounds of the geometry
+  check_bbox_geometry_match: true
   # check to see if there are too many links
   bloated_links: true
   # best practices - check for bloated metadata in properties

tests/test.config.yml

@@ -15,6 +15,8 @@ linting:
   check_unlocated: true
   # best practices - recommend items have a geometry
   check_geometry: true
+  # best practices - check if bbox matches the bounds of the geometry
+  check_bbox_geometry_match: true
   # check to see if there are too many links
   bloated_links: true
   # best practices - check for bloated metadata in properties

tests/test_lint.py

@@ -278,6 +278,71 @@ def test_unlocated_item():
     assert linter.check_geometry_null() == True
 
 
+def test_bbox_matches_geometry():
+    # Test with matching bbox and geometry
+    file = "sample_files/1.0.0/core-item.json"
+    linter = Linter(file)
+    assert linter.check_bbox_matches_geometry() == True
+
+    # Test with mismatched bbox and geometry
+    mismatched_item = {
+        "stac_version": "1.0.0",
+        "stac_extensions": [],
+        "type": "Feature",
+        "id": "test-item",
+        "bbox": [100.0, 0.0, 105.0, 1.0],  # Deliberately wrong bbox
+        "geometry": {
+            "type": "Polygon",
+            "coordinates": [
+                [
+                    [172.91173669923782, 1.3438851951615003],
+                    [172.95469614953714, 1.3438851951615003],
+                    [172.95469614953714, 1.3690476620161975],
+                    [172.91173669923782, 1.3690476620161975],
+                    [172.91173669923782, 1.3438851951615003],
+                ]
+            ],
+        },
+        "properties": {"datetime": "2020-12-11T22:38:32.125Z"},
+    }
+    linter = Linter(mismatched_item)
+    assert linter.check_bbox_matches_geometry() == False
+
+    # Test with null geometry (should return True as check is not applicable)
+    null_geom_item = {
+        "stac_version": "1.0.0",
+        "type": "Feature",
+        "id": "test-item-null-geom",
+        "bbox": [100.0, 0.0, 105.0, 1.0],
+        "geometry": None,
+        "properties": {"datetime": "2020-12-11T22:38:32.125Z"},
+    }
+    linter = Linter(null_geom_item)
+    assert linter.check_bbox_matches_geometry() == True
+
+    # Test with missing bbox (should return True as check is not applicable)
+    no_bbox_item = {
+        "stac_version": "1.0.0",
+        "type": "Feature",
+        "id": "test-item-no-bbox",
+        "geometry": {
+            "type": "Polygon",
+            "coordinates": [
+                [
+                    [172.91173669923782, 1.3438851951615003],
+                    [172.95469614953714, 1.3438851951615003],
+                    [172.95469614953714, 1.3690476620161975],
+                    [172.91173669923782, 1.3690476620161975],
+                    [172.91173669923782, 1.3438851951615003],
+                ]
+            ],
+        },
+        "properties": {"datetime": "2020-12-11T22:38:32.125Z"},
+    }
+    linter = Linter(no_bbox_item)
+    assert linter.check_bbox_matches_geometry() == True
+
+
 def test_bloated_item():
     file = "sample_files/1.0.0/core-item-bloated.json"
     linter = Linter(file)