Check for special files that might share names/extensions

Check for special files and hidden files that might be inadvertently
added to the zip bundle by the operating system in use that might
share names and/or extensions with the main bundle files.

1 files changed, 16 insertions, 1 deletions

diff --git a/r_qtl/r_qtl2.py b/r_qtl/r_qtl2.py
index 87491d0..0a96e7c 100644
--- a/r_qtl/r_qtl2.py
+++ b/r_qtl/r_qtl2.py
@@ -17,11 +17,26 @@ FILE_TYPES = (
     "geno", "founder_geno", "pheno", "covar", "phenocovar", "gmap", "pmap",
     "phenose")
 
+
+def __special_file__(filename):
+    """
+    Check whether the file is special in some ways, e.g. MacOSX seems to include
+    files in a directory `__MACOSX` that share parts of the name, and extensions
+    with the main files in the bundle.
+    """
+    is_macosx_special_file = filename.startswith("__MACOSX")
+    is_nix_hidden_file = Path(filename).name.startswith(".")
+
+    return (is_macosx_special_file or is_nix_hidden_file)
+
+
 def control_data(zfile: ZipFile) -> dict:
     """Retrieve the control file from the zip file info."""
     files = tuple(filename
                   for filename in zfile.namelist()
-                  if (filename.endswith(".yaml") or filename.endswith(".json")))
+                  if (not __special_file__(filename)
+                      and (filename.endswith(".yaml")
+                           or filename.endswith(".json"))))
     num_files = len(files)
     if num_files == 0:
         raise InvalidFormat("Expected a json or yaml control file.")