Merge pull request #32 from ssl-hep/feat/eos_pathfinder

Adding autocomplete URL resolution to eos paths for ATLAS CMS and Ope…

Author: ArturU043

servicex_analysis_utils/dataset_resolver.py

@@ -72,6 +72,19 @@ def ds_type_resolver(
     elif ds_name.startswith("root://") and ds_name.endswith("*"):
         return dataset.XRootD(ds_name)
 
+    elif ds_name.startswith("/eos/"):
+        if "/eos/opendata/" in ds_name:
+            return dataset.FileList([f"root://eospublic.cern.ch/{ds_name}"])
+        elif "/eos/atlas/" in ds_name:
+            return dataset.FileList([f"root://eosatlas.cern.ch/{ds_name}"])
+        elif "/eos/cms/" in ds_name:
+            return dataset.FileList([f"root://eoscms.cern.ch/{ds_name}"])
+        else:
+            raise ValueError(
+                f"Unable to determine the correct EOS instance for the provided path: {ds_name}."
+                "Please provide the full root:// URL. Cannot be a user path."
+            )
+
     elif re.match(r"^root://", ds_name):
         return dataset.FileList(ds_name)
 

tests/test_dataset_resolver.py

@@ -45,3 +45,18 @@
 def test_find_dataset(input_ds, expected_type):
     dataset = ds_type_resolver(input_ds)
     assert isinstance(dataset, expected_type)
+
+
+@pytest.mark.parametrize(
+    "eos_path, prefix",
+    [
+        ("/eos/opendata/atlas/rucio/somefile.root", "root://eospublic.cern.ch/"),
+        ("/eos/opendata/cms/rucio/somefile.root", "root://eospublic.cern.ch/"),
+        ("/eos/atlas/atlascerngroupdisk/somefile.root", "root://eosatlas.cern.ch/"),
+        ("/eos/cms/store/somefile.root", "root://eoscms.cern.ch/"),
+    ],
+)
+def test_eos_url_parsing(eos_path, prefix):
+    ds_out = ds_type_resolver(eos_path)
+    assert isinstance(ds_out, dataset.FileList)
+    assert ds_out.files[0] == prefix + eos_path