Merge pull request #134 from nansencenter/update_earthdata_cmr_provider

Update earthdata cmr provider

.github/workflows/ci.yml

@@ -20,7 +20,7 @@ jobs:
     env:
       BASE_IMAGE: "${{ vars.DOCKER_ORG }}/geospaas:2.5.2-python${{ matrix.python_version }}"
       IMAGE_NAME: "${{ vars.DOCKER_ORG }}/geospaas_harvesting"
-      METANORM_VERSION: '4.2.0'
+      METANORM_VERSION: '4.2.2'
       GEOSPAAS_DB_HOST: 'db'
       GEOSPAAS_DB_USER: 'test'
       GEOSPAAS_DB_PASSWORD: "${{ secrets.GEOSPAAS_DB_PASSWORD }}"

geospaas_harvesting/providers/earthdata_cmr.py

@@ -1,6 +1,7 @@
 """Code for searching EarthData CMR (https://www.earthdata.nasa.gov/)"""
 import json
 
+import shapely.errors
 from shapely.geometry import LineString, Point, Polygon
 
 import geospaas.catalog.managers as catalog_managers
@@ -18,7 +19,7 @@ def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
         self.search_url = 'https://cmr.earthdata.nasa.gov/search/granules.umm_json'
         self.search_parameters_parser.add_arguments([
-            WKTArgument('location', required=False, geometry_types=(LineString, Point, Polygon)),
+            EarthDataSpatialArgument('location', required=False, geometry_types=(LineString, Point, Polygon)),
             StringArgument('short_name', required=True, description='Short name of the collection'),
             ChoiceArgument('downloadable', valid_options=['true', 'false'], default='true'),
             StringArgument('platform'),
@@ -49,11 +50,34 @@ def _make_spatial_parameter(self, geometry):
             result = {'line': ','.join([f"{lon},{lat}" for lon, lat in points])}
         elif isinstance(geometry, Point):
             result = {'point': f"{geometry.xy[0][0]},{geometry.xy[1][0]}"}
+        elif isinstance(geometry, str):
+            name, value = geometry.split('=')
+            result = {name: value}
         else:
             raise ValueError(f"Unsupported geometry type {type(geometry)}")
         return result
 
 
+class EarthDataSpatialArgument(WKTArgument):
+    """Argument that provides the specific spatial format required by
+    queries to the CMR API.
+    See https://cmr.earthdata.nasa.gov/search/site/docs/search/api.html#g-spatial
+    for valid parameters.
+    """
+    def parse(self, value):
+        valid_prefixes = ('polygon', 'bounding_box', 'point', 'line', 'circle')
+        try:
+            return super().parse(value)
+        except (shapely.errors.ShapelyError, ValueError) as error:
+            if isinstance(value, str):
+                for prefix in valid_prefixes:
+                    if value.startswith(f"{prefix}=") or value.startswith(f"{prefix}[]="):
+                        return value
+            raise ValueError(
+                "location should be a geometry or a valid CMR spatial parameter"
+            ) from error
+
+
 class EarthDataCMRCrawler(HTTPPaginatedAPICrawler):
     """Crawler for the CMR Earthdata search API"""
 

tests/providers/test_earthdata_cmr.py

@@ -53,8 +53,39 @@ def test_make_spatial_parameter(self):
         self.assertEqual(
             provider._make_spatial_parameter(Point((1, 2))),
             {'point': '1.0,2.0'})
+        self.assertEqual(
+            provider._make_spatial_parameter('bounding_box=-180,60,180,90'),
+            {'bounding_box': '-180,60,180,90'})
         with self.assertRaises(ValueError):
             provider._make_spatial_parameter(MultiPoint(((1, 2), (3, 4))))
+        with self.assertRaises(ValueError):
+            provider._make_spatial_parameter('polygon:1.0,2.0,2.0,3.0,3.0,4.0,1.0,2.0')
+
+
+class EarthDataSpatialArgumentTestCase(unittest.TestCase):
+    """Tests for the Earthdata CMR spatial argument"""
+    def setUp(self):
+        self.argument = provider_earthdata_cmr.EarthDataSpatialArgument(
+            'location', geometry_types=(Polygon,))
+
+    def test_parse_wkt(self):
+        """Test parsing a WKT geometry"""
+        self.assertEqual(self.argument.parse('POLYGON((1 2,2 3,3 4,1 2))'),
+                         Polygon(((1, 2), (2, 3), (3, 4), (1, 2))))
+
+    def test_parse_raw_spatial_extent(self):
+        """Test parsing raw CMR spatial extent parameters"""
+        self.assertEqual(self.argument.parse('bounding_box=-180,60,180,90'),
+                         'bounding_box=-180,60,180,90')
+        self.assertEqual(self.argument.parse('polygon=1.0,2.0,2.0,3.0,3.0,4.0,1.0,2.0'),
+                         'polygon=1.0,2.0,2.0,3.0,3.0,4.0,1.0,2.0')
+        self.assertEqual(self.argument.parse('line=1.0,2.0,3.0,4.0'), 'line=1.0,2.0,3.0,4.0')
+        self.assertEqual(self.argument.parse('point=1.0,2.0'), 'point=1.0,2.0')
+
+        with self.assertRaises(ValueError):
+            self.argument.parse('foo=1.0,2.0')
+        with self.assertRaises(ValueError):
+            self.argument.parse('point:1.0,2.0')
 
 
 class EarthdataCMRCrawlerTestCase(unittest.TestCase):