Active interface for cf python: add active.do_active attribute and uri_analyzer() function

.gitignore

@@ -18,6 +18,7 @@ _sidebar.rst.inc
 
 # test output
 tests/test_data/cesm2_native_TREFHT.json
+tests/test_data/cesm2_native_cow_TREFHT.json
 tests/test_data/daily_data_tas.json
 tests/test_data/daily_data_ta.json
 tests/test_data/daily_data_masked_ta.json

activestorage/active.py

@@ -10,6 +10,44 @@
 from activestorage import netcdf_to_zarr as nz
 
 
+def uri_analyzer(uri):
+    """Run a first-pass examination on the file with given uri.
+
+    Parameters
+    ----------
+    uri: str
+        input ile URI.
+
+    Returns
+    -------
+    bool
+        Boolean that sets other methods/attributes in Active.
+
+    Raises
+    ------
+    ValueError
+        if file URI is None
+        if file URI is pointing to a nonexistent file
+    """
+    #TODO probably best placed in an acctive_tools library
+    result = True
+    if uri is None:
+        raise ValueError(f"Must use a valid file for uri. Got {uri}")
+    else:
+        if not os.path.isfile(uri):
+            raise ValueError(f"Must use existing file for uri. "
+                             f"{uri} not found")
+        else:
+            # bogus condition until we fix criteria
+            if "cow" in uri:
+                print("This file doesn't support active storage.")
+                result = False
+            else:
+                print("This file supports active storage.")
+
+        return result
+
+
 class Active:
     """ 
     Instantiates an interface to active storage which contains either zarr files
@@ -43,10 +81,15 @@ def __init__(self, uri, ncvar, missing_value=None, _FillValue=None, valid_min=No
         """
         # Assume NetCDF4 for now
         self.uri = uri
-        if self.uri is None:
-            raise ValueError(f"Must use a valid file for uri. Got {self.uri}")
-        if not os.path.isfile(self.uri):
-            raise ValueError(f"Must use existing file for uri. {self.uri} not found")
+        self.do_active = True
+        # run in active storage mode only if the uri analyzer returns True
+        uri_analyzer_result = uri_analyzer(self.uri)
+        self.do_active = uri_analyzer_result
+        # otherwise just get out and return active.do_active = False
+        if not self.do_active:
+            self._version = 1
+        else:
+            self._version = 2
         self.ncvar = ncvar
         if self.ncvar is None:
             raise ValueError("Must set a netCDF variable name to slice")

tests/unit/test_active.py

@@ -3,7 +3,7 @@
 import pytest
 import threading
 
-from activestorage.active import Active
+from activestorage.active import Active, uri_analyzer
 
 
 def test_uri_none():
@@ -105,3 +105,30 @@ def test_lock():
 
     active.lock = None
     assert active.lock is False
+
+
+def test_uri_analyzer_invalid_uri():
+    """Test the uri_analyzer functionality."""
+    #TODO implement correct test cases when fully functional
+    uri = "tests/test_data/cesm2_native_cow.nc"
+    ncvar = "TREFHT"
+    active = Active(uri, ncvar=ncvar)
+    assert active.do_active is False
+    raised_exc = "'Active' object has no attribute '_ncvar'"
+
+    # test version switching
+    item = active.__getitem__(index=3)[0, 0]
+    expected = np.array(277.11945, dtype="float32")
+    np.testing.assert_array_equal(item, expected)
+
+
+def test_uri_analyzer_valid_uri():
+    """Test the uri_analyzer functionality."""
+    #TODO implement correct test cases when fully functional
+    uri = "tests/test_data/cesm2_native.nc"
+    ncvar = "TREFHT"
+    active = Active(uri, ncvar=ncvar)
+    assert active.do_active is True
+
+    active._method = "min"
+    assert active.method([3,444]) == 3