added pre-filters to expand_user_input private function, removed unnecessary test cases

Author: yucongalicechen

src/diffpy/labpdfproc/labpdfprocapp.py

@@ -3,7 +3,6 @@
 
 from diffpy.labpdfproc.functions import apply_corr, compute_cve
 from diffpy.labpdfproc.tools import (
-    expand_list_file,
     known_sources,
     load_user_metadata,
     set_input_lists,
@@ -29,14 +28,9 @@ def get_args(override_cli_inputs=None):
         "'.' (load everything in the current directory), 'data' (load"
         "everything in the folder ./data), 'data/file_list.txt' (load"
         " the list of files contained in the text-file called "
-        "file_list.txt that can be found in the folder ./data). "
-        "\nWildcard character (*) is accepted. Examples include './*.chi'"
-        " (load all files with .chi extension), 'data/*.chi' (load all "
-        "files in 'data' file with .chi extension), 'file*.chi' (load all "
-        "files starting with 'file' and ending with .chi extension), 'test*' "
-        "(load all files and directories starting with 'test'), 'test*/*.chi' "
-        "(load all directories starting with 'test' and all files under "
-        "with .chi extension). ",
+        "file_list.txt that can be found in the folder ./data), "
+        "'data/*.chi' (load all files with extension .chi in the "
+        "folder ./data), 'data*' (load all files and directories starting with 'data').",
     )
     p.add_argument(
         "-a",
@@ -102,7 +96,6 @@ def get_args(override_cli_inputs=None):
 
 def main():
     args = get_args()
-    args = expand_list_file(args)
     args = set_input_lists(args)
     args.output_directory = set_output_directory(args)
     args.wavelength = set_wavelength(args)
@@ -126,7 +119,7 @@ def main():
             )
 
         input_pattern = Diffraction_object(wavelength=args.wavelength)
-        xarray, yarray = loadData(args.input_file, unpack=True)
+        xarray, yarray = loadData(filepath, unpack=True)
         input_pattern.insert_scattering_quantity(
             xarray,
             yarray,

src/diffpy/labpdfproc/tests/test_tools.py

@@ -6,8 +6,6 @@
 
 from diffpy.labpdfproc.labpdfprocapp import get_args
 from diffpy.labpdfproc.tools import (
-    expand_list_file,
-    expand_wildcard_file,
     known_sources,
     load_user_metadata,
     set_input_lists,
@@ -81,10 +79,6 @@
         ["unreadable*.txt"],
         ["unreadable_file.txt"],
     ),
-    (  # wildcard pattern, matching directories starting with input and all files under with .chi extension
-        ["input*/*.chi"],
-        ["input_dir/good_data.chi"],
-    ),
 ]
 
 
@@ -96,8 +90,6 @@ def test_set_input_lists(inputs, expected, user_filesystem):
 
     cli_inputs = ["2.5"] + inputs
     actual_args = get_args(cli_inputs)
-    actual_args = expand_wildcard_file(actual_args)
-    actual_args = expand_list_file(actual_args)
     actual_args = set_input_lists(actual_args)
     assert sorted(actual_args.input_paths) == sorted(expected_paths)
 
@@ -121,16 +113,6 @@ def test_set_input_lists(inputs, expected, user_filesystem):
         ["input_dir/file_list.txt"],
         "Cannot find missing_file.txt. Please specify valid input file(s) or directories.",
     ),
-    (  # valid wildcard pattern, but does not match any files or directories
-        ["non_existing_dir*"],
-        "Invalid wildcard input non_existing_dir*. "
-        "Please ensure the wildcard pattern matches at least one file or directory.",
-    ),
-    (  # invalid wildcard pattern
-        ["invalid_dir**"],
-        "Invalid wildcard input invalid_dir**. "
-        "Please ensure the wildcard pattern matches at least one file or directory.",
-    ),
 ]
 
 
@@ -141,8 +123,6 @@ def test_set_input_files_bad(inputs, msg, user_filesystem):
     cli_inputs = ["2.5"] + inputs
     actual_args = get_args(cli_inputs)
     with pytest.raises(FileNotFoundError, match=msg[0]):
-        actual_args = expand_wildcard_file(actual_args)
-        actual_args = expand_list_file(actual_args)
         actual_args = set_input_lists(actual_args)
 
 

src/diffpy/labpdfproc/tools.py

@@ -1,4 +1,3 @@
-import glob
 from pathlib import Path
 
 WAVELENGTHS = {"Mo": 0.71, "Ag": 0.59, "Cu": 1.54}
@@ -29,9 +28,9 @@ def set_output_directory(args):
     return output_dir
 
 
-def expand_wildcard_file(args):
+def _expand_user_input(args):
     """
-    Expands wildcard inputs by adding all files or directories within directories matching the pattern.
+    Expands the list of inputs by adding files from file lists and wildcards.
 
     Parameters
     ----------
@@ -40,45 +39,17 @@ def expand_wildcard_file(args):
 
     Returns
     -------
-    the arguments with the wildcard inputs expanded
+    the arguments with the modified input list
 
     """
     wildcard_inputs = [input_name for input_name in args.input if "*" in input_name]
+    file_list_inputs = [input_name for input_name in args.input if "file_list" in input_name]
+
     for wildcard_input in wildcard_inputs:
-        if not glob.glob(wildcard_input):
-            raise FileNotFoundError(
-                f"Invalid wildcard input {wildcard_input}. "
-                f"Please ensure the wildcard pattern matches at least one file or directory."
-            )
-        input_files = Path(".").glob(wildcard_input)
-        for input_file in input_files:
-            if input_file.is_file():
-                args.input.append(str(input_file))
-            elif input_file.is_dir():
-                files = input_file.glob("*")
-                inputs = [str(file) for file in files if file.is_file() and "file_list" not in file.name]
-                args.input.extend(inputs)
-            else:
-                raise FileNotFoundError(f"Invalid wildcard input {wildcard_input}.")
+        input_files = [str(file) for file in Path(".").glob(wildcard_input)]
+        args.input.extend(input_files)
         args.input.remove(wildcard_input)
-    return args
 
-
-def expand_list_file(args):
-    """
-    Expands the list of inputs by adding files from file lists and removing the file list.
-
-    Parameters
-    ----------
-    args argparse.Namespace
-        the arguments from the parser
-
-    Returns
-    -------
-    the arguments with the modified input list
-
-    """
-    file_list_inputs = [input_name for input_name in args.input if "file_list" in input_name]
     for file_list_input in file_list_inputs:
         with open(file_list_input, "r") as f:
             file_inputs = [input_name.strip() for input_name in f.readlines()]
@@ -106,6 +77,7 @@ def set_input_lists(args):
     """
 
     input_paths = []
+    args = _expand_user_input(args)
     for input_name in args.input:
         input_path = Path(input_name).resolve()
         if input_path.exists():