Support Downloading series memeber with diff names

Author: push2stack

py_src/fusion/utils.py

@@ -112,6 +112,8 @@
     "tar",
 ]
 
+FILENAME_CLEAN_RE = re.compile(r"[^a-zA-Z0-9_\-]")
+
 re_str_1 = re.compile("(.)([A-Z][a-z]+)")
 re_str_2 = re.compile("([a-z0-9])([A-Z])")
 
@@ -568,6 +570,7 @@ def distribution_to_filename(
     sep = "/"
     if "\\" in root_folder:
         sep = "\\"
+    final_name = _clean_filename(final_name)
     return f"{root_folder}{sep}{final_name}"
 
 
@@ -764,6 +767,15 @@ def path_to_url(x: str, is_raw: bool = False, is_download: bool = False) -> str:
     return "/".join(distribution_to_url("", dataset, date, ext, catalog, is_download).split("/")[1:])
 
 
+def _clean_filename(filename: str) -> str:
+    path = Path(filename)
+    stem = path.stem
+    ext = path.suffix
+    new_stem = FILENAME_CLEAN_RE.sub("_", stem)
+    new_base = new_stem + ext
+    return new_base
+
+
 def upload_files(  # noqa: PLR0913
     fs_fusion: fsspec.AbstractFileSystem,
     fs_local: fsspec.AbstractFileSystem,

py_tests/test_utils.py

@@ -20,6 +20,7 @@
 from fusion.fusion import Fusion
 from fusion.utils import (
     PathLikeT,
+    _clean_filename,
     _filename_to_distribution,
     _merge_responses,
     convert_date_format,
@@ -523,6 +524,11 @@ def test_path_to_url() -> None:
     )
 
 
+def test_clean_filename() -> None:
+    cleaned = _clean_filename("bad name:report.csv")
+    assert cleaned == "bad_name_report.csv"
+
+
 def test_filename_to_distribution() -> None:
     file_name = "dataset__catalog__datasetseries.csv"
     catalog, dataset, datasetseries, file_format = _filename_to_distribution(file_name)