pyproject.toml: remove stale comment from dependency (#10546)

dvcfs.get: simplify and separate files into respective filesystems while collecting/walking

dvc/dependency/repo.py

@@ -101,9 +101,9 @@ class RepoDependency(Dependency):
             return
 
         hashes: list[tuple[str, HashInfo, dict[str, Any]]] = []
-        for src_path, dest_path, *rest in files:
+        for src_path, dest_path, maybe_info in files:
             try:
-                info = rest[0] if rest else self.fs.info(src_path)
+                info = maybe_info or self.fs.info(src_path)
                 hash_info = info["dvc_info"]["entry"].hash_info
                 dest_info = to.fs.info(dest_path)
             except (KeyError, AttributeError):

dvc/fs/__init__.py

@@ -1,5 +1,6 @@
 import glob
-from typing import Optional, Union
+from itertools import repeat
+from typing import Optional
 from urllib.parse import urlparse
 
 from dvc.config import ConfigError as RepoConfigError
@@ -47,7 +48,7 @@ known_implementations.update(
 
 def download(
     fs: "FileSystem", fs_path: str, to: str, jobs: Optional[int] = None
-) -> list[Union[tuple[str, str], tuple[str, str, dict]]]:
+) -> list[tuple[str, str, Optional[dict]]]:
     from dvc.scm import lfs_prefetch
 
     from .callbacks import TqdmCallback
@@ -84,7 +85,7 @@ def download(
         cb.set_size(len(from_infos))
         jobs = jobs or fs.jobs
         generic.copy(fs, from_infos, localfs, to_infos, callback=cb, batch_size=jobs)
-        return list(zip(from_infos, to_infos))
+        return list(zip(from_infos, to_infos, repeat(None)))
 
 
 def parse_external_url(url, fs_config=None, config=None):

dvc/fs/dvc.py

@@ -4,8 +4,8 @@ import ntpath
 import os
 import posixpath
 import threading
-from collections import deque
-from contextlib import ExitStack, suppress
+from collections import defaultdict, deque
+from contextlib import ExitStack, nullcontext, suppress
 from glob import has_magic
 from typing import TYPE_CHECKING, Any, Callable, Optional, Union
 
@@ -19,6 +19,8 @@ from dvc_objects.fs.base import AnyFSPath, FileSystem
 from .data import DataFileSystem
 
 if TYPE_CHECKING:
+    from contextlib import AbstractContextManager
+
     from dvc.repo import Repo
     from dvc.types import DictStrAny, StrPath
 
@@ -498,7 +500,7 @@ class _DVCFileSystem(AbstractFileSystem):
             **kwargs,
         )
 
-    def _get(  # noqa: C901
+    def _get(  # noqa: C901, PLR0912
         self,
         rpath,
         lpath,
@@ -507,7 +509,7 @@ class _DVCFileSystem(AbstractFileSystem):
         maxdepth=None,
         batch_size=None,
         **kwargs,
-    ) -> list[Union[tuple[str, str], tuple[str, str, dict]]]:
+    ) -> list[tuple[str, str, Optional[dict]]]:
         if (
             isinstance(rpath, list)
             or isinstance(lpath, list)
@@ -531,10 +533,13 @@ class _DVCFileSystem(AbstractFileSystem):
         if self.isfile(rpath):
             with callback.branched(rpath, lpath) as child:
                 self.get_file(rpath, lpath, callback=child, **kwargs)
-                return [(rpath, lpath)]
+                return [(rpath, lpath, None)]
 
-        _files = []
+        result: list[tuple[str, str, Optional[dict]]] = []
         _dirs: list[str] = []
+        _files: dict[FileSystem, list[tuple[str, str, Optional[dict]]]]
+        _files = defaultdict(list)
+
         for root, dirs, files in self.walk(rpath, maxdepth=maxdepth, detail=True):
             if files:
                 callback.set_size((callback.size or 0) + len(files))
@@ -550,33 +555,47 @@ class _DVCFileSystem(AbstractFileSystem):
             _, dvc_fs, _ = self._get_subrepo_info(key)
 
             for name, info in files.items():
+                dvc_info = info.get("dvc_info")
+                fs_info = info.get("fs_info")
+                if dvc_fs and dvc_info and not fs_info:
+                    fs = dvc_fs
+                    fs_path = dvc_info["name"]
+                else:
+                    fs = self.repo.fs
+                    fs_path = fs_info["name"]
+
                 src_path = f"{root}{self.sep}{name}"
                 dest_path = f"{dest_root}{os.path.sep}{name}"
-                _files.append((dvc_fs, src_path, dest_path, info))
+                _files[fs].append((fs_path, dest_path, dvc_info))
+                result.append((src_path, dest_path, info))
 
         os.makedirs(lpath, exist_ok=True)
         for d in _dirs:
             os.mkdir(d)
 
-        def _get_file(arg):
-            dvc_fs, src, dest, info = arg
-            dvc_info = info.get("dvc_info")
-            if dvc_info and dvc_fs:
-                dvc_path = dvc_info["name"]
-                dvc_fs.get_file(
-                    dvc_path, dest, callback=callback, info=dvc_info, **kwargs
-                )
-            else:
-                self.get_file(src, dest, callback=callback, **kwargs)
-            return src, dest, info
+        def get_file(arg: tuple[FileSystem, tuple[str, str, Optional[dict]]]):
+            fs, (src, dest, info) = arg
+            kw = kwargs
+            if isinstance(fs, DataFileSystem):
+                kw = kw | {"info": info}
+            return fs.get_file(src, dest, callback=callback, **kw)
 
-        with ThreadPoolExecutor(max_workers=batch_size) as executor:
-            return list(executor.imap_unordered(_get_file, _files))
+        if batch_size == 1:
+            ctx: AbstractContextManager = nullcontext()
+            map_fn: Callable = map
+        else:
+            ctx = ThreadPoolExecutor(max_workers=batch_size)
+            map_fn = ctx.imap_unordered
+
+        with ctx:
+            it = ((fs, f) for fs, files in _files.items() for f in files)
+            deque(map_fn(get_file, it), maxlen=0)
+        return result
 
     def get_file(self, rpath, lpath, **kwargs):
+        dvc_info = kwargs.pop("info", {}).pop("dvc_info", None)
         key = self._get_key_from_relative(rpath)
         fs_path = self._from_key(key)
-
         dirpath = os.path.dirname(lpath)
         if dirpath:
             # makedirs raises error if the string is empty
@@ -590,7 +609,7 @@ class _DVCFileSystem(AbstractFileSystem):
                 raise
 
         dvc_path = _get_dvc_path(dvc_fs, subkey)
-        return dvc_fs.get_file(dvc_path, lpath, **kwargs)
+        return dvc_fs.get_file(dvc_path, lpath, info=dvc_info, **kwargs)
 
     def du(self, path, total=True, maxdepth=None, withdirs=False, **kwargs):
         if maxdepth is not None:
@@ -666,7 +685,7 @@ class DVCFileSystem(FileSystem):
         recursive: bool = False,
         batch_size: Optional[int] = None,
         **kwargs,
-    ) -> list[Union[tuple[str, str], tuple[str, str, dict]]]:
+    ) -> list[tuple[str, str, Optional[dict]]]:
         # FileSystem.get is non-recursive by default if arguments are lists
         # otherwise, it's recursive.
         recursive = not (isinstance(from_info, list) and isinstance(to_info, list))

pyproject.toml

@@ -45,7 +45,7 @@ dependencies = [
     "dvc-studio-client>=0.21,<1",
     "dvc-task>=0.3.0,<1",
     "flatten_dict<1,>=0.4.1",
-    "flufl.lock>=8.1.0,<9", # https://github.com/iterative/dvc/issues/9654
+    "flufl.lock>=8.1.0,<9",
     "fsspec>=2024.2.0",
     "funcy>=1.14",
     "grandalf<1,>=0.7",