33  FS

Python functions similar to R’s {fs}

import pandas as pd
from siuba import _, group_by, mutate, select, filter
import fs

33.1 List Directory

from pathlib import Path

def dir_ls(path):
    return [p for p in Path(path).iterdir()]


dir_ls("d")

[PosixPath('d/test1.txt'), PosixPath('d/e2'), PosixPath('d/e1')]

33.1.1 Custom Implement fs.dir_ls()

d_paths = fs.dir_ls("d", recurse=True)
d_paths

[PosixPath('d/test1.txt'),
 PosixPath('d/e2'),
 PosixPath('d/e2/test3.txt'),
 PosixPath('d/e2/test3-cp.txt'),
 PosixPath('d/e1'),
 PosixPath('d/e1/test2.txt'),
 PosixPath('d/e1/test2-cp.txt'),
 PosixPath('d/e1/test2-copy.txt')]

d_paths_df = pd.DataFrame({"paths": d_paths})
d_paths_df

	paths
0	d/test1.txt
1	d/e2
2	d/e2/test3.txt
3	d/e2/test3-cp.txt
4	d/e1
5	d/e1/test2.txt
6	d/e1/test2-cp.txt
7	d/e1/test2-copy.txt

33.2 File Name

from pathlib import Path
from typing import Union, List

def path_file(path: Union[str, Path, List[str], List[Path]]) -> Union[Path, List[Path]]:
    """Get filename(s) from path(s).

    Parameters
    ----------
    path : str, Path, or list of str/Path
        Single path or list of paths

    Returns
    -------
    Path or list of Path
        Filename(s) as PosixPath object(s)
    """
    if isinstance(path, (str, Path)):
        return Path(Path(path).name)
    return [Path(Path(p).name) for p in path]

print(d_paths_df['paths'][0])
path_file(d_paths_df['paths'][0])

d/test1.txt

PosixPath('test1.txt')

d_paths_df.assign(file_name = path_file(d_paths_df["paths"]))

	paths	file_name
0	d/test1.txt	test1.txt
1	d/e2	e2
2	d/e2/test3.txt	test3.txt
3	d/e2/test3-cp.txt	test3-cp.txt
4	d/e1	e1
5	d/e1/test2.txt	test2.txt
6	d/e1/test2-cp.txt	test2-cp.txt
7	d/e1/test2-copy.txt	test2-copy.txt

33.3 Copy file

import shutil
from pathlib import Path

def file_copy(src, dst):
    shutil.copy2(src, dst)

# Example usage:
file_copy("d/e1/test2.txt", "d/e1/test2-copy.txt")

33.3.1 Custom fs.file_copy()

fs.file_copy(path = ["d/e1/test2.txt", "d/e2/test3.txt"],
             new_path = ["d/e1/test2-cp.txt", "d/e2/test3-cp.txt"],
             overwrite = True)

fs.dir_ls("d", recurse=True)

[PosixPath('d/test1.txt'),
 PosixPath('d/e2'),
 PosixPath('d/e2/test3.txt'),
 PosixPath('d/e2/test3-cp.txt'),
 PosixPath('d/e1'),
 PosixPath('d/e1/test2.txt'),
 PosixPath('d/e1/test2-cp.txt'),
 PosixPath('d/e1/test2-copy.txt')]

33.4 Copy Folder

import shutil
from pathlib import Path

def dir_copy(src, dst):
    shutil.copytree(src, dst)

d_paths_df2 = d_paths_df.assign(is_file = d_paths_df["paths"].map(str).str.contains("\\.txt"))
d_paths_df2

	paths	is_file
0	d/test1.txt	True
1	d/e2	False
2	d/e2/test3.txt	True
3	d/e2/test3-cp.txt	True
4	d/e1	False
5	d/e1/test2.txt	True
6	d/e1/test2-cp.txt	True
7	d/e1/test2-copy.txt	True

d_paths_df3 = (d_paths_df2
               >> filter(_.is_file == False)
               >> mutate(new_paths=_.paths.map(str) + "_cp"))

d_paths_df3

	paths	is_file	new_paths
1	d/e2	False	d/e2_cp
4	d/e1	False	d/e1_cp

33.4.1 Custom fs.dir_copy()

fs.dir_copy(d_paths_df3["paths"], d_paths_df3["new_paths"])