Function for cleaning up mothur output directory based on current files and an optional whitelist.

tidy_mothur_output.py

def tidy_output_dir(mothur_obj, whitelist=list()):
    """
    Removes files from the output directory that are not listed in the current_files attribute or a whitelist.
    
    USE WITH CAUTION: This will delete any files in the mothur objects output directory (which may just be the current working
    directory) which can have unintended consequences.
    
    This function was breifly included in mothur-py.utils, but the danger of its accidental use deleting important files has
    relegated it to a gist.
    
    """

    # append current files to whitelist
    for file in mothur_obj.current_files.values():
        whitelist.append(file)

    # tidy up environment to help save space
    out_dir = mothur_obj.current_dirs.get('output', None)

    # purge non-whitelist files from the output directory
    for file in os.listdir(out_dir):

        # conditionally correct filepath
        if out_dir is not None:
            file = os.path.join(out_dir, file)

        # only process files, not directories
        if os.path.isfile(file):
            if file not in whitelist:
                os.remove(file)