refactor kit path/glob utils

try to use pathlib over os

WrightTools/kit/_glob.py

@@ -6,7 +6,7 @@
 from .._open import open
 
 
-__all__ = ["describe_wt5", "filter_wt5s", "glob_wt5s", "search_for_attrs",]
+__all__ = ["describe_wt5", "filter_wt5s", "glob_handler", "glob_wt5s", "search_for_attrs",]
 
 
 def describe_wt5(path: Union[str, PathLike]) -> dict:
@@ -37,6 +37,30 @@ def glob_wt5s(directory: Union[str, PathLike], recursive=True) -> Iterator:
     return pathlib.Path(directory).glob(pattern)
 
 
+def glob_handler(extension, folder=None, identifier=None, recursive=True) -> List[pathlib.Path]:
+    """Return a list of all files matching specified inputs.
+
+    Parameters
+    ----------
+    extension : string
+        File extension.
+    folder : string (optional)
+        Folder to search within. Default is None (current working
+        directory).
+    identifier : string
+        Unique identifier. Default is None.
+    recursive : bool
+        When true, searches folder and all subfolders for identifier
+
+    Returns
+    -------
+    list of pathlib.Path objects
+        path objects for matching files.
+    """
+    pattern = f"**/*.{extension}" if recursive else f"*.{extension}"
+    return [x for x in filter(lambda x: identifier in str(x), pathlib.Path(folder).glob(pattern))]
+
+
 def search_for_attrs(
         directory: Union[str, PathLike],
         recursive=True,
@@ -63,7 +87,7 @@ def search_for_attrs(
     
     Example
     -------
-    To find a scan based on scan parameters in the bluesky-cmds:
+    To find a scan from scan parameters in the bluesky-cmds:
     >>> search_wt5_by_attr(os.environ["WT5_DATA_DIR"], name="primary", shape=[136,101], )
     """
     return filter_wt5s(glob_wt5s(directory, recursive), **kwargs)

WrightTools/kit/_path.py

@@ -3,78 +3,41 @@
 # --- import --------------------------------------------------------------------------------------
 
 
-import os
-import glob
+import pathlib
 
 
 # --- define --------------------------------------------------------------------------------------
 
 
-__all__ = ["get_path_matching", "glob_handler"]
+__all__ = ["get_path_matching"]
 
 
 # --- functions -----------------------------------------------------------------------------------
 
 
-def get_path_matching(name):
-    """Get path matching a name.
+def get_path_matching(name:str) -> pathlib.Path:
+    """
+    Non-recursive search for path to the folder "name".
+    Searches the user directory, then looks up the cwd for a parent folder that matches.
 
     Parameters
     ----------
     name : string
-        Name to search for.
+        name of directory to search for.
 
     Returns
     -------
-    string
-        Full filepath.
+    pathlib.Path
+        Full filepath to directory name.
+
     """
     # first try looking in the user folder
-    p = os.path.join(os.path.expanduser("~"), name)
+    p = pathlib.Path.home() / name
     # then try expanding upwards from cwd
-    if not os.path.isdir(p):
+    if not p.is_dir():
         p = None
-        drive, folders = os.path.splitdrive(os.getcwd())
-        folders = folders.split(os.sep)
-        folders.insert(0, os.sep)
+        drive, *folders = pathlib.Path.cwd().parts
         if name in folders:
-            p = os.path.join(drive, *folders[: folders.index(name) + 1])
+            p = pathlib.Path(drive).joinpath(*folders[:folders.index(name)+1])
     # TODO: something more robust to catch the rest of the cases?
     return p
-
-
-def glob_handler(extension, folder=None, identifier=None):
-    """Return a list of all files matching specified inputs.
-
-    Parameters
-    ----------
-    extension : string
-        File extension.
-    folder : string (optional)
-        Folder to search within. Default is None (current working
-        directory).
-    identifier : string
-        Unique identifier. Default is None.
-
-    Returns
-    -------
-    list of strings
-        Full path of matching files.
-    """
-    filepaths = []
-    if folder:
-        # comment out [ and ]...
-        folder = folder.replace("[", "?")
-        folder = folder.replace("]", "*")
-        folder = folder.replace("?", "[[]")
-        folder = folder.replace("*", "[]]")
-        glob_str = os.path.join(folder, "*" + extension)
-    else:
-        glob_str = "*" + extension + "*"
-    for filepath in glob.glob(glob_str):
-        if identifier:
-            if identifier in filepath:
-                filepaths.append(filepath)
-        else:
-            filepaths.append(filepath)
-    return filepaths