added functionality of handling an input of a file list

Author: yucongalicechen

src/diffpy/labpdfproc/labpdfprocapp.py

@@ -21,7 +21,7 @@ def get_args(override_cli_inputs=None):
         "data-files in that directory will be processed. Examples of valid "
         "inputs are 'file.xy', 'data/file.xy', 'file.xy, data/file.xy', "
         "'.' (load everything in the current directory), 'data' (load"
-        "everything in the folder ./data', 'data/file_list.txt' (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).",
     )

src/diffpy/labpdfproc/tests/test_tools.py

@@ -27,6 +27,9 @@
     (  # glob input directory
         ["./input_dir"],
         [
+            "./good_data.chi",
+            "./good_data.xy",
+            "./good_data.txt",
             "input_dir/good_data.chi",
             "input_dir/good_data.xy",
             "input_dir/good_data.txt",
@@ -69,7 +72,7 @@ def test_set_input_lists(inputs, expected, user_filesystem):
     cli_inputs = ["2.5"] + inputs
     actual_args = get_args(cli_inputs)
     actual_args = set_input_lists(actual_args)
-    assert list(actual_args.input_paths).sort() == expected_paths.sort()
+    assert set(actual_args.input_paths) == set(expected_paths)
 
 
 # This test covers non-existing single input file or directory, in this case we raise an error with message

src/diffpy/labpdfproc/tools.py

@@ -28,6 +28,24 @@ def set_output_directory(args):
     return output_dir
 
 
+def _parse_file_list_file(input_path):
+    with open(input_path, "r") as f:
+        lines = [line.strip() for line in f]
+        input_files = [Path(line).resolve() for line in lines if Path(line).is_file()]
+        return input_files
+
+
+def _parse_input_paths(input_path):
+    # Takes a path to return either a list of files paths if it is a file list,
+    # a list of single file path if it is a data file, or nothing
+    if "file_list" in input_path.name:
+        return _parse_file_list_file(input_path)
+    elif input_path.is_file():
+        return [input_path]
+    else:
+        return []
+
+
 def set_input_lists(args):
     """
     Set input directory and files.
@@ -51,16 +69,16 @@ def set_input_lists(args):
         input_path = Path(input).resolve()
         if input_path.exists():
             if input_path.is_file():
-                input_paths.append(input_path)
+                input_paths.extend(_parse_input_paths(input_path))
             elif input_path.is_dir():
                 input_files = input_path.glob("*")
-                input_files = [file.resolve() for file in input_files if file.is_file()]
-                input_paths.extend(input_files)
+                for file in input_files:
+                    input_paths.extend(_parse_input_paths(file))
             else:
                 raise FileNotFoundError(f"Cannot find {input}. Please specify valid input file(s) or directories.")
         else:
             raise FileNotFoundError(f"Cannot find {input}")
-    setattr(args, "input_paths", input_paths)
+    setattr(args, "input_paths", list(set(input_paths)))
     return args