address #966: better error messages for non-gzip files with gz ext

Author: jeromedockes

nibabel/loadsave.py

@@ -20,6 +20,47 @@
 from .deprecated import deprecate_with_version
 
 
+def _signature_matches_extension(filename):
+    """Check if signature aka magic number matches filename extension.
+
+    Parameters
+    ----------
+    filename : str or os.PathLike
+        Path to the file to check
+
+    Returns
+    -------
+    matches : bool
+       - `True` if the filename extension is not recognized (not .gz nor .bz2)
+       - `True` if the magic number was successfully read and corresponds to
+         the format indicated by the extension.
+       - `False` otherwise.
+    error_message : str
+       An error message if opening the file failed or a mismatch is detected;
+       the empty string otherwise.
+
+    """
+    signatures = {
+        ".gz": {"signature": b"\x1f\x8b", "format_name": "gzip"},
+        ".bz2": {"signature": b"\x42\x5a\x68", "format_name": "bzip2"}
+    }
+    filename = _stringify_path(filename)
+    *_, ext = splitext_addext(filename)
+    ext = ext.lower()
+    if ext not in signatures:
+        return True, ""
+    expected_signature = signatures[ext]["signature"]
+    try:
+        with open(filename, "rb") as fh:
+            found_signature = fh.read(len(expected_signature))
+    except Exception:
+        return False, f"Could not read file: {filename}"
+    if found_signature == expected_signature:
+        return True, ""
+    format_name = signatures[ext]["format_name"]
+    return False, f"File {filename} is not a {format_name} file"
+
+
 def load(filename, **kwargs):
     r""" Load file given filename, guessing at file type
 
@@ -44,6 +85,9 @@ def load(filename, **kwargs):
         raise FileNotFoundError(f"No such file or no access: '{filename}'")
     if stat_result.st_size <= 0:
         raise ImageFileError(f"Empty file: '{filename}'")
+    matches, msg = _signature_matches_extension(filename)
+    if not matches:
+        raise ImageFileError(msg)
 
     sniff = None
     for image_klass in all_image_classes:

nibabel/tests/test_loadsave.py

@@ -76,6 +76,21 @@ def test_load_empty_image():
     assert str(err.value).startswith('Empty file: ')
 
 
+@pytest.mark.parametrize("extension", [".gz", ".bz2"])
+def test_load_bad_compressed_extension(tmp_path, extension):
+    file_path = tmp_path.joinpath(f"img.nii{extension}")
+    file_path.write_bytes(b"bad")
+    with pytest.raises(ImageFileError, match=".*is not a .* file"):
+        load(file_path)
+
+
+def test_load_file_that_cannot_be_read(tmp_path):
+    subdir = tmp_path.joinpath("img.nii.gz")
+    subdir.mkdir()
+    with pytest.raises(ImageFileError, match="Could not read"):
+        load(subdir)
+
+
 def test_read_img_data_nifti():
     shape = (2, 3, 4)
     data = np.random.normal(size=shape)