test: use paramiko for ssh instead of sshfs (#1014)

* use paramiko instead of sshfs

* use specified port

* test default handler behaviour

* default to fsspec instead of error if scheme not found

* attempt to close socket

* fix ci

* Revert "fix ci"

This reverts commit e56e337.

* broader exception

* also handle socket exception

* get user robust

pyproject.toml

@@ -64,7 +64,7 @@ test = [
   "fsspec",
   "fsspec-xrootd",
   "s3fs; python_version<\"3.12\"",  # asyncio not available
-  "sshfs; python_version<\"3.12\"",  # asyncio not available
+  "paramiko",
   "pytest>=6",
   "pytest-timeout",
   "pytest-rerunfailures",

src/uproot/_util.py

@@ -421,13 +421,14 @@ def file_path_to_source_class(file_path, options):
         elif _windows_absolute_path_pattern_slash.match(parsed_url_path) is not None:
             windows_absolute_path = parsed_url_path[1:]
 
+    scheme = parsed_url.scheme.lower()
     if (
-        parsed_url.scheme.upper() == "FILE"
+        scheme == "file"
         or len(parsed_url.scheme) == 0
         or windows_absolute_path is not None
     ):
         if windows_absolute_path is None:
-            if parsed_url.netloc.upper() == "LOCALHOST":
+            if parsed_url.netloc.lower() == "localhost":
                 file_path = parsed_url_path
             else:
                 file_path = parsed_url.netloc + parsed_url_path
@@ -458,7 +459,7 @@ def file_path_to_source_class(file_path, options):
             )
         return out, os.path.expanduser(file_path)
 
-    elif parsed_url.scheme.upper() == "ROOT":
+    elif scheme == "root":
         out = options["xrootd_handler"]
         if out is None:
             out = uproot.source.xrootd.XRootDSource
@@ -482,7 +483,7 @@ def file_path_to_source_class(file_path, options):
             )
         return out, file_path
 
-    elif parsed_url.scheme.upper() in {"S3"}:
+    elif scheme == "s3":
         out = options["s3_handler"]
         if out is None:
             out = uproot.source.s3.S3Source
@@ -505,7 +506,7 @@ def file_path_to_source_class(file_path, options):
             )
         return out, file_path
 
-    elif parsed_url.scheme.upper() in {"HTTP", "HTTPS"}:
+    elif scheme in ("http", "https"):
         out = options["http_handler"]
         if out is None:
             out = uproot.source.http.HTTPSource
@@ -530,6 +531,10 @@ def file_path_to_source_class(file_path, options):
         return out, file_path
 
     else:
+        # try to use fsspec before raising an error
+        if scheme in _schemes:
+            return uproot.source.fsspec.FSSpecSource, file_path
+
         raise ValueError(f"URI scheme not recognized: {file_path}")
 
 

tests/test_0692_fsspec.py

@@ -1,6 +1,5 @@
 # BSD 3-Clause License; see https://github.com/scikit-hep/uproot4/blob/main/LICENSE
 
-import fsspec
 import pytest
 import uproot
 import uproot.source.fsspec
@@ -67,38 +66,38 @@ def test_open_fsspec_s3(handler):
         assert len(data) == 8004
 
 
-def test_open_fsspec_ssh():
-    pytest.importorskip("sshfs")
-
-    # check localhost has ssh access to itself
-    try:
-        user = subprocess.check_output(["whoami"]).strip().decode("ascii")
-        host = "localhost"
-        ssh_command = ["ssh", f"{user}@{host}", "'echo hello'"]
-        result = subprocess.run(
-            ssh_command,
-            shell=True,
-            text=True,
-            capture_output=True,
-        )
-        assert (
-            result.returncode == 0
-        ), f"ssh access to localhost failed with {result.stderr}"
-    except Exception as e:
-        pytest.skip(f"ssh access to localhost failed with {e}")
+@pytest.mark.parametrize(
+    "handler",
+    [
+        uproot.source.fsspec.FSSpecSource,
+        None,
+    ],
+)
+def test_open_fsspec_ssh(handler):
+    pytest.importorskip("paramiko")
+    import paramiko
+    import getpass
 
-    # at this time sshfs does not implement cat_file. This will alert us if it ever does
-    with pytest.raises(NotImplementedError):
-        fs = fsspec.filesystem("ssh", host="localhost")
-        fs.cat_file("some-file", start=0, end=100)
+    user = getpass.getuser()
+    host = "localhost"
+    port = 22
 
-    pytest.skip("sshfs does not implement cat_file")
+    # only test this if we can connect to the host (this will work in GitHub Actions)
+    try:
+        with paramiko.SSHClient() as client:
+            client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
+            client.connect(hostname=host, port=port, username=user)
+    except (
+        paramiko.ssh_exception.SSHException,
+        paramiko.ssh_exception.NoValidConnectionsError,
+    ) as e:
+        pytest.skip(f"ssh connection to host failed: {e}")
 
     # cache the file
     local_path = skhep_testdata.data_path("uproot-issue121.root")
 
-    uri = f"ssh://{user}@{host}:22{local_path}"
-    with uproot.open(uri, handler=uproot.source.fsspec.FSSpecSource) as f:
+    uri = f"ssh://{user}@{host}:{port}{local_path}"
+    with uproot.open(uri, handler=handler) as f:
         data = f["Events/MET_pt"].array(library="np")
         assert len(data) == 40