lfs: drop dvc-objects dependency

Author: pmrowla

pyproject.toml

@@ -25,13 +25,12 @@ dependencies = [
     "dulwich>=0.21.6",
     "pygit2>=1.14.0",
     "pygtrie>=2.3.2",
-    "fsspec>=2024.2.0",
+    "fsspec[tqdm]>=2024.2.0",
     "pathspec>=0.9.0",
     "asyncssh>=2.13.1,<3",
     "funcy>=1.14",
-    "shortuuid>=0.5.0",
-    "dvc-objects>=4,<5",
     "aiohttp-retry>=2.5.0",
+    "tqdm",
 ]
 
 [project.urls]
@@ -54,6 +53,7 @@ tests = [
     "types-certifi==2021.10.8.3",
     "types-mock==5.1.0.2",
     "types-paramiko==3.4.0.20240120",
+    "types-tqdm",
 ]
 dev = [
     "scmrepo[tests]",

src/scmrepo/git/lfs/client.py

@@ -1,13 +1,14 @@
 import logging
-from collections.abc import Iterable
-from contextlib import AbstractContextManager
+import os
+import shutil
+from collections.abc import Iterable, Iterator
+from contextlib import AbstractContextManager, contextmanager, suppress
 from multiprocessing import cpu_count
+from tempfile import NamedTemporaryFile
 from typing import TYPE_CHECKING, Any, Optional
 
 import aiohttp
 from aiohttp_retry import ExponentialRetry, RetryClient
-from dvc_objects.fs import localfs
-from dvc_objects.fs.utils import as_atomic
 from fsspec.asyn import _run_coros_in_chunks, sync_wrapper
 from fsspec.callbacks import DEFAULT_CALLBACK
 from fsspec.implementations.http import HTTPFileSystem
@@ -156,11 +157,11 @@ async def _download(
         **kwargs,
     ):
         async def _get_one(from_path: str, to_path: str, **kwargs):
-            with as_atomic(localfs, to_path, create_parents=True) as tmp_file:
+            with _as_atomic(to_path, create_parents=True) as tmp_file:
                 with callback.branched(from_path, tmp_file) as child:
                     await self._fs._get_file(
                         from_path, tmp_file, callback=child, **kwargs
-                    )  # pylint: disable=protected-access
+                    )
                     callback.relative_update()
 
         resp_data = await self._batch_request(objects, **kwargs)
@@ -184,3 +185,21 @@ async def _get_one(from_path: str, to_path: str, **kwargs):
                 raise result
 
     download = sync_wrapper(_download)
+
+
+@contextmanager
+def _as_atomic(to_info: str, create_parents: bool = False) -> Iterator[str]:
+    parent = os.path.dirname(to_info)
+    if create_parents:
+        os.makedirs(parent, exist_ok=True)
+
+    tmp_file = NamedTemporaryFile(dir=parent, delete=False)
+    tmp_file.close()
+    try:
+        yield tmp_file.name
+    except BaseException:
+        with suppress(FileNotFoundError):
+            os.unlink(tmp_file.name)
+        raise
+    else:
+        shutil.move(tmp_file.name, to_info)

src/scmrepo/git/lfs/progress.py

@@ -1,11 +1,114 @@
-from typing import BinaryIO, Callable, Optional, Union
+import logging
+import sys
+from typing import Any, BinaryIO, Callable, ClassVar, Optional, Union
 
-from dvc_objects.fs.callbacks import TqdmCallback
-from fsspec.callbacks import DEFAULT_CALLBACK, Callback
+from fsspec.callbacks import DEFAULT_CALLBACK, Callback, TqdmCallback
+from tqdm import tqdm
 
 from scmrepo.progress import GitProgressEvent
 
 
+class _Tqdm(tqdm):
+    """
+    maximum-compatibility tqdm-based progressbars
+    """
+
+    BAR_FMT_DEFAULT = (
+        "{percentage:3.0f}% {desc}|{bar}|"
+        "{postfix[info]}{n_fmt}/{total_fmt}"
+        " [{elapsed}<{remaining}, {rate_fmt:>11}]"
+    )
+    # nested bars should have fixed bar widths to align nicely
+    BAR_FMT_DEFAULT_NESTED = (
+        "{percentage:3.0f}%|{bar:10}|{desc:{ncols_desc}.{ncols_desc}}"
+        "{postfix[info]}{n_fmt}/{total_fmt}"
+        " [{elapsed}<{remaining}, {rate_fmt:>11}]"
+    )
+    BAR_FMT_NOTOTAL = "{desc}{bar:b}|{postfix[info]}{n_fmt} [{elapsed}, {rate_fmt:>11}]"
+    BYTES_DEFAULTS: ClassVar[dict[str, Any]] = {
+        "unit": "B",
+        "unit_scale": True,
+        "unit_divisor": 1024,
+        "miniters": 1,
+    }
+
+    def __init__(  # noqa: PLR0913
+        self,
+        iterable=None,
+        disable=None,
+        level=logging.ERROR,
+        desc=None,
+        leave=False,
+        bar_format=None,
+        bytes=False,  # noqa: A002
+        file=None,
+        total=None,
+        postfix=None,
+        **kwargs,
+    ):
+        kwargs = kwargs.copy()
+        if bytes:
+            kwargs = {**self.BYTES_DEFAULTS, **kwargs}
+        else:
+            kwargs.setdefault("unit_scale", total > 999 if total else True)
+        if file is None:
+            file = sys.stderr
+        super().__init__(
+            iterable=iterable,
+            disable=disable,
+            leave=leave,
+            desc=desc,
+            bar_format="!",
+            lock_args=(False,),
+            total=total,
+            **kwargs,
+        )
+        self.postfix = postfix or {"info": ""}
+        if bar_format is None:
+            if self.__len__():
+                self.bar_format = (
+                    self.BAR_FMT_DEFAULT_NESTED if self.pos else self.BAR_FMT_DEFAULT
+                )
+            else:
+                self.bar_format = self.BAR_FMT_NOTOTAL
+        else:
+            self.bar_format = bar_format
+        self.refresh()
+
+    def update_to(self, current, total=None):
+        if total:
+            self.total = total
+        self.update(current - self.n)
+
+    def close(self):
+        self.postfix["info"] = ""
+        # remove ETA (either unknown or zero); remove completed bar
+        self.bar_format = self.bar_format.replace("<{remaining}", "").replace(
+            "|{bar:10}|", " "
+        )
+        super().close()
+
+    @property
+    def format_dict(self):
+        """inject `ncols_desc` to fill the display width (`ncols`)"""
+        d = super().format_dict
+        ncols = d["ncols"] or 80
+        # assumes `bar_format` has max one of ("ncols_desc" & "ncols_info")
+
+        meter = self.format_meter(  # type: ignore[call-arg]
+            ncols_desc=1, ncols_info=1, **d
+        )
+        ncols_left = ncols - len(meter) + 1
+        ncols_left = max(ncols_left, 0)
+        if ncols_left:
+            d["ncols_desc"] = d["ncols_info"] = ncols_left
+        else:
+            # work-around for zero-width description
+            d["ncols_desc"] = d["ncols_info"] = 1
+            d["prefix"] = ""
+        return d
+
+
 class LFSCallback(Callback):
     """Callback subclass to generate Git/LFS style progress."""
 
@@ -37,7 +140,11 @@ def _update_git(self):
     def branched(self, path_1: Union[str, BinaryIO], path_2: str, **kwargs):
         if self.git_progress:
             return TqdmCallback(
-                bytes=True, desc=path_1 if isinstance(path_1, str) else path_2
+                tqdm_kwargs={
+                    "desc": path_1 if isinstance(path_1, str) else path_2,
+                    "bytes": True,
+                },
+                tqdm_cls=_Tqdm,
             )
         return DEFAULT_CALLBACK