use token_urlsafe instead of shortuuid to generate temp file (#269)

Python>=3.6 has secrets.token_urlsafe that can create cryptographically
safe strings/numbers. So we don't need to use shortuuid.
The generated text string is of the same size as `shortuuid.uuid()`,
but can contain underscore and hyphen characters (not just alpha-numeric values),
which might make it less human-readable. But I am not sure if it's worth it to use
a third-party library (which can be gradually removed).

Author: skshetry

pyproject.toml

@@ -22,7 +22,6 @@ classifiers = [
 requires-python = ">=3.8"
 dynamic = ["version"]
 dependencies = [
-    "shortuuid>=0.5.0",
     "funcy>=1.14",
     "fsspec>=2022.10.0",
 ]

src/dvc_objects/fs/utils.py

@@ -7,6 +7,7 @@
 import threading
 from concurrent import futures
 from contextlib import contextmanager, suppress
+from secrets import token_urlsafe
 from typing import TYPE_CHECKING, Any, Collection, Dict, Iterator, Optional, Set, Union
 
 from dvc_objects.executors import ThreadPoolExecutor
@@ -60,10 +61,8 @@ def move(src: "AnyFSPath", dst: "AnyFSPath") -> None:
     case src and dst are on different filesystems and actual physical copying
     of data is happening.
     """
-    from shortuuid import uuid
-
     dst = os.path.abspath(dst)
-    tmp = f"{dst}.{uuid()}"
+    tmp = tmp_fname(dst)
 
     if os.path.islink(src):
         shutil.copy(src, tmp)
@@ -173,11 +172,9 @@ def copyfile(
         shutil.copyfileobj(fsrc, wrapped, length=LOCAL_CHUNK_SIZE)
 
 
-def tmp_fname(fname: "AnyFSPath" = "") -> "AnyFSPath":
+def tmp_fname(prefix: str = "") -> str:
     """Temporary name for a partial download"""
-    from shortuuid import uuid
-
-    return os.fspath(fname) + "." + uuid() + ".tmp"
+    return f"{prefix}.{token_urlsafe(16)}.tmp"
 
 
 @contextmanager

tests/fs/test_utils.py

@@ -10,7 +10,7 @@ def test_tmp_fname():
     file = os.path.join("path", "to", "file")
 
     def pattern(path):
-        return r"^" + re.escape(path) + r"\.[a-z0-9]{22}\.tmp$"
+        return r"^" + re.escape(path) + r"\.[a-z0-9_-]{22}\.tmp$"
 
     assert re.search(pattern(file), utils.tmp_fname(file), re.IGNORECASE)
     assert re.search(