Only use xarray engine for grib data

src/earthkit/data/readers/grib/__init__.py

@@ -25,6 +25,19 @@ def _is_default(magic, content_type):
     return (magic is None or len(magic) == 0) and (content_type is None or len(content_type) == 0)
 
 
+def is_grib_file(path):
+    from earthkit.data.core.settings import SETTINGS
+
+    n_bytes = SETTINGS.get("reader-type-check-bytes")
+    magic = None
+    with open(path, "rb") as f:
+        magic = f.read(n_bytes)
+
+    if _match_magic(magic, False):
+        return True
+    return _match_magic(magic, True)
+
+
 def reader(source, path, *, magic=None, deeper_check=False, **kwargs):
     if _match_magic(magic, deeper_check):
         from .file import GRIBReader

src/earthkit/data/utils/xarray/engine.py

@@ -293,23 +293,32 @@ def open_dataset(
 
     @classmethod
     def guess_can_open(cls, filename_or_obj):
-        return True  # filename_or_obj.endswith(".grib")
+        try:
+            from earthkit.data.core import Base
+
+            if isinstance(filename_or_obj, Base):
+                return True
+            elif isinstance(filename_or_obj, str):
+                from earthkit.data.readers.grib import is_grib_file
+
+                return is_grib_file(filename_or_obj)
+        except Exception:
+            LOG.debug(
+                "Failed to guess if %s can be opened by the earthkit backend", filename_or_obj, exc_info=True
+            )
+
+        return False
 
     @staticmethod
     def _fieldlist(filename_or_obj, source_type):
         from earthkit.data.core import Base
 
         if isinstance(filename_or_obj, Base):
             ds = filename_or_obj
-        # TODO: Add Path? or handle with try statement
         elif isinstance(filename_or_obj, str):
             from earthkit.data import from_source
 
             ds = from_source(source_type, filename_or_obj)
-        else:
-            from earthkit.data import from_object
-
-            ds = from_object(filename_or_obj)
         return ds
 
 

tests/readers/test_geotiff_reader.py

@@ -32,6 +32,16 @@ def test_geotiff_reader_with_multiband():
     assert s[0].shape == (294, 315)
 
 
+@pytest.mark.skipif(NO_RIOXARRAY, reason="rioxarray not available")
+@pytest.mark.with_proj
+def test_geotiff_bypassing_xr_engine():
+    import xarray as xr
+
+    # the earthkit xarray engine should not try to handle this file but leave it to the rioxarray backend
+    da = xr.open_dataarray(earthkit_test_data_file("dgm50hs_col_32_368_5616_nw.tif"))
+    assert da.name == "band_data"
+
+
 if __name__ == "__main__":
     from earthkit.data.testing import main