rimport: Print msg if file copied to staging but not relinked.

Author: samsrabin

rimport

@@ -178,15 +178,9 @@ def stage_data(
                 f"({staging_root})"
             )
         print(f"{INDENT}File is already published and linked.")
-        if can_file_be_downloaded(src.resolve(), staging_root):
-            print(f"{INDENT}File is available for download.")
-        else:
-            print(f"{INDENT}File is not (yet) available for download.")
-        return
-
-    # TODO: Check whether file is published but not relinked.
-    if check:
-        print(f"{INDENT}File is not already published")
+        print_can_file_be_downloaded(
+            can_file_be_downloaded(src.resolve(), staging_root)
+        )
         return
 
     if not src.exists():
@@ -204,6 +198,20 @@ def stage_data(
         ) from exc
 
     dst = staging_root / rel
+
+    if dst.exists():
+        print(f"{INDENT}File is already published but NOT linked; do")
+        print(f"{2*INDENT}relink.py {rel}")
+        print(f"{INDENT}to resolve.")
+        print_can_file_be_downloaded(
+            can_file_be_downloaded(rel, staging_root)
+        )
+        return
+
+    if check:
+        print(f"{INDENT}File is not already published")
+        return
+
     dst.parent.mkdir(parents=True, exist_ok=True)
     shutil.copy2(src, dst)
     print(f"{INDENT}[rimport] staged {src} -> {dst}")
@@ -295,6 +303,18 @@ def can_file_be_downloaded(file_relpath: Path, staging_root: Path, timeout: floa
         return False
 
 
+def print_can_file_be_downloaded(file_can_be_downloaded: bool):
+    """Print a message indicating whether a file is available for download.
+
+    Args:
+        file_can_be_downloaded: Boolean indicating if the file can be downloaded.
+    """
+    if file_can_be_downloaded:
+        print(f"{INDENT}File is available for download.")
+    else:
+        print(f"{INDENT}File is not (yet) available for download.")
+
+
 def main(argv: List[str] | None = None) -> int:
     """Main entry point for the rimport tool.
 

tests/rimport/test_stage_data.py

@@ -154,6 +154,39 @@ def test_prints_live_symlink_already_published_is_downloadable(
         msg = "is already under staging directory"
         assert msg not in stdout
 
+    def test_prints_published_but_not_linked(
+        self, inputdata_root, staging_root, capsys
+    ):
+        """
+        Tests printed message for when a file has been published (copied to staging root) but not
+        yet linked (inputdata version replaced with symlink to staging version).
+        """
+        # Create a real file in staging AND in inputdata
+        filename = "real_file.nc"
+        staged = staging_root / filename
+        staged.write_text("data")
+        inputdata = inputdata_root / filename
+        inputdata.write_text("data")
+
+        # Mock shutil.copy2 to verify it's never called
+        with patch("shutil.copy2") as mock_copy:
+            # Mock can_file_be_downloaded to return True
+            with patch("rimport.can_file_be_downloaded", return_value=True):
+                # Should print message for live symlink and return early
+                rimport.stage_data(inputdata, inputdata_root, staging_root)
+
+                # Verify that shutil.copy2 was never called (function returned early)
+                mock_copy.assert_not_called()
+
+        # Verify the right messages were printed or not
+        stdout = capsys.readouterr().out.strip()
+        msg = "File is already published and linked"
+        assert msg not in stdout
+        msg = "File is already published but NOT linked; do"
+        assert msg in stdout
+        msg = "File is available for download"
+        assert msg in stdout
+
     def test_raises_error_for_live_symlink_pointing_somewhere_other_than_staging(
         self, tmp_path, inputdata_root, staging_root
     ):