Merge pull request #59 from yucongalicechen/input_dir4

implement wildcard pattern for input

Author: sbillinge

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,
@@ -29,7 +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).",
+        "file_list.txt that can be found in the folder ./data), "
+        "'./*.chi', 'data/*.chi' (load all files with extension .chi in the "
+        "folder ./data).",
     )
     p.add_argument(
         "-a",
@@ -95,13 +96,12 @@ 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)
     args = load_user_metadata(args)
 
-    for filepath in args.input_directory:
+    for filepath in args.input_paths:
         outfilestem = filepath.stem + "_corrected"
         corrfilestem = filepath.stem + "_cve"
         outfile = args.output_directory / (outfilestem + ".chi")
@@ -119,13 +119,13 @@ 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,
             "tth",
             scat_quantity="x-ray",
-            name=str(args.input_file),
+            name=filepath.stem,
             metadata={"muD": args.mud, "anode_type": args.anode_type},
         )
 

src/diffpy/labpdfproc/tests/conftest.py

@@ -39,6 +39,7 @@ def user_filesystem(tmp_path):
     with open(input_dir / "file_list.txt", "w") as f:
         f.write("good_data.chi \n good_data.xy \n good_data.txt \n missing_file.txt")
     with open(input_dir / "file_list_example2.txt", "w") as f:
+        f.write("input_dir/*.txt \n")
         f.write("input_dir/good_data.chi \n")
         f.write("good_data.xy \n")
         f.write(f"{str(input_dir.resolve() / 'good_data.txt')}\n")

src/diffpy/labpdfproc/tests/test_tools.py

@@ -6,7 +6,6 @@
 
 from diffpy.labpdfproc.labpdfprocapp import get_args
 from diffpy.labpdfproc.tools import (
-    expand_list_file,
     known_sources,
     load_user_metadata,
     set_input_lists,
@@ -50,9 +49,21 @@
             "input_dir/binary.pkl",
         ],
     ),
-    (  # file_list_example2.txt list of files provided in different directories
+    (  # file_list_example2.txt list of files provided in different directories with wildcard
         ["input_dir/file_list_example2.txt"],
-        ["input_dir/good_data.chi", "good_data.xy", "input_dir/good_data.txt"],
+        ["input_dir/good_data.chi", "good_data.xy", "input_dir/good_data.txt", "input_dir/unreadable_file.txt"],
+    ),
+    (  # wildcard pattern, matching files with .chi extension in the same directory
+        ["./*.chi"],
+        ["good_data.chi"],
+    ),
+    (  # wildcard pattern, matching files with .chi extension in the input directory
+        ["input_dir/*.chi"],
+        ["input_dir/good_data.chi"],
+    ),
+    (  # wildcard pattern, matching files starting with good_data
+        ["good_data*"],
+        ["good_data.chi", "good_data.xy", "good_data.txt"],
     ),
 ]
 
@@ -65,7 +76,6 @@ def test_set_input_lists(inputs, expected, user_filesystem):
 
     cli_inputs = ["2.5"] + inputs
     actual_args = get_args(cli_inputs)
-    actual_args = expand_list_file(actual_args)
     actual_args = set_input_lists(actual_args)
     assert sorted(actual_args.input_paths) == sorted(expected_paths)
 
@@ -98,7 +108,6 @@ def test_set_input_files_bad(inputs, msg, user_filesystem):
     os.chdir(base_dir)
     cli_inputs = ["2.5"] + inputs
     actual_args = get_args(cli_inputs)
-    actual_args = expand_list_file(actual_args)
     with pytest.raises(FileNotFoundError, match=msg[0]):
         actual_args = set_input_lists(actual_args)
 

src/diffpy/labpdfproc/tools.py

@@ -28,9 +28,9 @@ def set_output_directory(args):
     return output_dir
 
 
-def expand_list_file(args):
+def _expand_user_input(args):
     """
-    Expands the list of inputs by adding files from file lists and removing the file list.
+    Expands the list of inputs by adding files from file lists and wildcards.
 
     Parameters
     ----------
@@ -48,6 +48,11 @@ def expand_list_file(args):
             file_inputs = [input_name.strip() for input_name in f.readlines()]
         args.input.extend(file_inputs)
         args.input.remove(file_list_input)
+    wildcard_inputs = [input_name for input_name in args.input if "*" in input_name]
+    for wildcard_input in wildcard_inputs:
+        input_files = [str(file) for file in Path(".").glob(wildcard_input) if "file_list" not in file.name]
+        args.input.extend(input_files)
+        args.input.remove(wildcard_input)
     return args
 
 
@@ -70,6 +75,7 @@ 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():
@@ -86,7 +92,7 @@ def set_input_lists(args):
                     f"Cannot find {input_name}. Please specify valid input file(s) or directories."
                 )
         else:
-            raise FileNotFoundError(f"Cannot find {input_name}")
+            raise FileNotFoundError(f"Cannot find {input_name}.")
     setattr(args, "input_paths", list(set(input_paths)))
     return args