fix: refactor data server protocols to be lazy

Author: jasonb5

CIME/Servers/__init__.py

@@ -1,19 +1,74 @@
-# pylint: disable=import-error
+"""
+CIME Server implementations for data transfer.
+
+Server availability is detected lazily on first access to avoid
+running executables at import time.
+"""
+
+from functools import lru_cache
 from shutil import which
 
-has_gftp = which("globus-url-copy")
-has_svn = which("svn")
-has_wget = which("wget")
-has_ftp = True
-try:
-    from ftplib import FTP
-except ImportError:
-    has_ftp = False
-if has_ftp:
-    from CIME.Servers.ftp import FTP
-if has_svn:
-    from CIME.Servers.svn import SVN
-if has_wget:
-    from CIME.Servers.wget import WGET
-if has_gftp:
-    from CIME.Servers.gftp import GridFTP
+
+@lru_cache(maxsize=None)
+def is_protocol_available(protocol: str) -> bool:
+    """
+    Check if a protocol is available.
+
+    Args:
+        protocol: One of 'ftp', 'svn', 'wget', 'gftp'
+
+    Returns:
+        True if the protocol is available, False otherwise.
+    """
+    protocol = protocol.lower()
+    if protocol == "ftp":
+        try:
+            from ftplib import FTP  # noqa: F401 pylint: disable=unused-import
+
+            return True
+        except ImportError:
+            return False
+    elif protocol == "svn":
+        return which("svn") is not None
+    elif protocol == "wget":
+        return which("wget") is not None
+    elif protocol == "gftp":
+        return which("globus-url-copy") is not None
+    return False
+
+
+def __getattr__(name: str):
+    """Lazy loading of server classes and has_* attributes."""
+    if name == "FTP":
+        if is_protocol_available("ftp"):
+            from CIME.Servers.ftp import FTP
+
+            return FTP
+        raise AttributeError("FTP server not available")
+    elif name == "SVN":
+        if is_protocol_available("svn"):
+            from CIME.Servers.svn import SVN
+
+            return SVN
+        raise AttributeError("SVN server not available (svn not found)")
+    elif name == "WGET":
+        if is_protocol_available("wget"):
+            from CIME.Servers.wget import WGET
+
+            return WGET
+        raise AttributeError("WGET server not available (wget not found)")
+    elif name == "GridFTP":
+        if is_protocol_available("gftp"):
+            from CIME.Servers.gftp import GridFTP
+
+            return GridFTP
+        raise AttributeError("GridFTP server not available (globus-url-copy not found)")
+    elif name == "has_ftp":
+        return is_protocol_available("ftp")
+    elif name == "has_svn":
+        return is_protocol_available("svn")
+    elif name == "has_wget":
+        return is_protocol_available("wget")
+    elif name == "has_gftp":
+        return is_protocol_available("gftp")
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

CIME/case/check_input_data.py

@@ -24,7 +24,7 @@ def _download_checksum_file(rundir):
     # download and merge all available chksum files.
     while protocol is not None:
         protocol, address, user, passwd, chksum_file, _, _ = inputdata.get_next_server()
-        if protocol not in vars(CIME.Servers):
+        if not CIME.Servers.is_protocol_available(protocol):
             logger.info("Client protocol {} not enabled".format(protocol))
             continue
         logger.info(
@@ -414,7 +414,7 @@ def _check_input_data_impl(
     no_files_missing = True
     server = None
     if download:
-        if protocol not in vars(CIME.Servers):
+        if not CIME.Servers.is_protocol_available(protocol):
             logger.info("Client protocol {} not enabled".format(protocol))
             return False
         logger.info(

CIME/tests/test_unit_servers.py

@@ -0,0 +1,117 @@
+"""Unit tests for CIME.Servers lazy loading.
+
+These tests verify:
+1. Lazy loading behavior (no subprocess at import time)
+2. Backward compatibility with existing usage patterns
+3. is_protocol_available() API
+"""
+
+import pytest
+
+
+class TestServersLazyLoading:
+    """Tests for lazy loading of server modules."""
+
+    def test_import_does_not_run_which(self):
+        """Importing CIME.Servers should not run shutil.which()."""
+        import CIME.Servers
+
+        assert CIME.Servers is not None
+
+    def test_availability_not_checked_at_import(self):
+        """Availability flags should be unchecked after import."""
+        import importlib
+        import sys
+
+        # Force reimport
+        for mod in list(sys.modules.keys()):
+            if "CIME.Servers" in mod:
+                del sys.modules[mod]
+
+        import CIME.Servers
+
+        assert CIME.Servers is not None
+
+
+class TestProtocolAvailability:
+    """Tests for is_protocol_available() API."""
+
+    def test_is_protocol_available_ftp(self):
+        """FTP should always be available (stdlib)."""
+        import CIME.Servers
+
+        assert CIME.Servers.is_protocol_available("ftp") is True
+
+    def test_is_protocol_available_invalid(self):
+        """Invalid protocol should return False."""
+        import CIME.Servers
+
+        assert CIME.Servers.is_protocol_available("invalid") is False
+
+    def test_is_protocol_available_case_insensitive(self):
+        """Protocol check should be case-insensitive."""
+        import CIME.Servers
+
+        assert CIME.Servers.is_protocol_available("FTP") is True
+        assert CIME.Servers.is_protocol_available("Ftp") is True
+        assert CIME.Servers.is_protocol_available("ftp") is True
+
+
+class TestServerClassAccess:
+    """Tests for accessing server classes."""
+
+    def test_ftp_attribute_access(self):
+        """Accessing CIME.Servers.FTP should work via __getattr__."""
+        import CIME.Servers
+
+        FTP = CIME.Servers.FTP
+        assert FTP is not None
+        assert FTP.__name__ == "FTP"
+
+    def test_ftp_has_ftp_login_method(self):
+        """FTP class should have ftp_login class method (used by check_input_data)."""
+        import CIME.Servers
+
+        assert hasattr(CIME.Servers.FTP, "ftp_login")
+
+    def test_wget_has_wget_login_method(self):
+        """WGET class should have wget_login class method if available."""
+        import CIME.Servers
+
+        if CIME.Servers.is_protocol_available("wget"):
+            assert hasattr(CIME.Servers.WGET, "wget_login")
+
+    def test_unavailable_server_raises_attribute_error(self):
+        """Accessing unavailable server should raise AttributeError."""
+        import CIME.Servers
+
+        if not CIME.Servers.is_protocol_available("gftp"):
+            with pytest.raises(AttributeError):
+                _ = CIME.Servers.GridFTP
+
+
+class TestBackwardCompatibility:
+    """Tests for backward compatibility with existing code patterns."""
+
+    def test_has_ftp_attribute(self):
+        """has_ftp attribute should return True (FTP always available)."""
+        import CIME.Servers
+
+        assert CIME.Servers.has_ftp is True
+
+    def test_has_attributes_exist(self):
+        """All has_* attributes should be accessible."""
+        import CIME.Servers
+
+        # These should not raise, regardless of availability
+        _ = CIME.Servers.has_ftp
+        _ = CIME.Servers.has_svn
+        _ = CIME.Servers.has_wget
+        _ = CIME.Servers.has_gftp
+
+    def test_instantiate_ftp_server(self):
+        """Should be able to instantiate FTP server (backward compat)."""
+        import CIME.Servers
+
+        FTP = CIME.Servers.FTP
+        assert FTP is not None