huggingface-hub 1.0.0rc1__tar.gz → 1.0.0rc3__tar.gz

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: huggingface_hub
-Version: 1.0.0rc1
+Version: 1.0.0rc3
 Summary: Client library to download and publish models, datasets and other repos on the huggingface.co hub
 Home-page: https://github.com/huggingface/huggingface_hub
 Author: Hugging Face, Inc.
@@ -105,7 +105,7 @@ If you prefer, you can also install it with [conda](https://huggingface.co/docs/
 In order to keep the package minimal by default, `huggingface_hub` comes with optional dependencies useful for some use cases. For example, if you want have a complete experience for Inference, run:
 
 ```bash
-pip install huggingface_hub[inference]
+pip install "huggingface_hub[inference]"
 ```
 
 To learn more installation and optional dependencies, check out the [installation guide](https://huggingface.co/docs/huggingface_hub/en/installation).

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/README.md

@@ -65,7 +65,7 @@ If you prefer, you can also install it with [conda](https://huggingface.co/docs/
 In order to keep the package minimal by default, `huggingface_hub` comes with optional dependencies useful for some use cases. For example, if you want have a complete experience for Inference, run:
 
 ```bash
-pip install huggingface_hub[inference]
+pip install "huggingface_hub[inference]"
 ```
 
 To learn more installation and optional dependencies, check out the [installation guide](https://huggingface.co/docs/huggingface_hub/en/installation).

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/setup.py

@@ -131,7 +131,6 @@ setup(
     extras_require=extras,
     entry_points={
         "console_scripts": [
-            "huggingface-cli=huggingface_hub.commands.huggingface_cli:main",
             "hf=huggingface_hub.cli.hf:main",
             "tiny-agents=huggingface_hub.inference._mcp.cli:app",
         ],

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/src/huggingface_hub/__init__.py

@@ -46,7 +46,7 @@ import sys
 from typing import TYPE_CHECKING
 
 
-__version__ = "1.0.0.rc1"
+__version__ = "1.0.0.rc3"
 
 # Alphabetical order of definitions is ensured in tests
 # WARNING: any comment added in this dictionary definition will be lost when
@@ -138,6 +138,7 @@ _SUBMOD_ATTRS = {
         "push_to_hub_fastai",
     ],
     "file_download": [
+        "DryRunFileInfo",
         "HfFileMetadata",
         "_CACHED_NO_EXIST",
         "get_hf_file_metadata",
@@ -513,8 +514,6 @@ _SUBMOD_ATTRS = {
         "CorruptedCacheException",
         "DeleteCacheStrategy",
         "HFCacheInfo",
-        "HfHubAsyncTransport",
-        "HfHubTransport",
         "cached_assets_path",
         "close_session",
         "dump_environment_info",
@@ -625,6 +624,7 @@ __all__ = [
     "DocumentQuestionAnsweringInputData",
     "DocumentQuestionAnsweringOutputElement",
     "DocumentQuestionAnsweringParameters",
+    "DryRunFileInfo",
     "EvalResult",
     "FLAX_WEIGHTS_NAME",
     "FeatureExtractionInput",
@@ -645,8 +645,6 @@ __all__ = [
     "HfFileSystemFile",
     "HfFileSystemResolvedPath",
     "HfFileSystemStreamFile",
-    "HfHubAsyncTransport",
-    "HfHubTransport",
     "ImageClassificationInput",
     "ImageClassificationOutputElement",
     "ImageClassificationOutputTransform",
@@ -1147,6 +1145,7 @@ if TYPE_CHECKING:  # pragma: no cover
     )
     from .file_download import (
         _CACHED_NO_EXIST,  # noqa: F401
+        DryRunFileInfo,  # noqa: F401
         HfFileMetadata,  # noqa: F401
         get_hf_file_metadata,  # noqa: F401
         hf_hub_download,  # noqa: F401
@@ -1515,8 +1514,6 @@ if TYPE_CHECKING:  # pragma: no cover
         CorruptedCacheException,  # noqa: F401
         DeleteCacheStrategy,  # noqa: F401
         HFCacheInfo,  # noqa: F401
-        HfHubAsyncTransport,  # noqa: F401
-        HfHubTransport,  # noqa: F401
         cached_assets_path,  # noqa: F401
         close_session,  # noqa: F401
         dump_environment_info,  # noqa: F401

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/src/huggingface_hub/_commit_api.py

@@ -33,6 +33,7 @@ from .utils import (
     validate_hf_hub_args,
 )
 from .utils import tqdm as hf_tqdm
+from .utils._runtime import is_xet_available
 
 
 if TYPE_CHECKING:
@@ -353,7 +354,7 @@ def _warn_on_overwriting_operations(operations: list[CommitOperation]) -> None:
 
 
 @validate_hf_hub_args
-def _upload_lfs_files(
+def _upload_files(
     *,
     additions: list[CommitOperationAdd],
     repo_type: str,
@@ -362,6 +363,86 @@ def _upload_lfs_files(
     endpoint: Optional[str] = None,
     num_threads: int = 5,
     revision: Optional[str] = None,
+    create_pr: Optional[bool] = None,
+):
+    """
+    Negotiates per-file transfer (LFS vs Xet) and uploads in batches.
+    """
+    xet_additions: list[CommitOperationAdd] = []
+    lfs_actions: list[dict[str, Any]] = []
+    lfs_oid2addop: dict[str, CommitOperationAdd] = {}
+
+    for chunk in chunk_iterable(additions, chunk_size=UPLOAD_BATCH_MAX_NUM_FILES):
+        chunk_list = [op for op in chunk]
+
+        transfers: list[str] = ["basic", "multipart"]
+        has_buffered_io_data = any(isinstance(op.path_or_fileobj, io.BufferedIOBase) for op in chunk_list)
+        if is_xet_available():
+            if not has_buffered_io_data:
+                transfers.append("xet")
+            else:
+                logger.warning(
+                    "Uploading files as a binary IO buffer is not supported by Xet Storage. "
+                    "Falling back to HTTP upload."
+                )
+
+        actions_chunk, errors_chunk, chosen_transfer = post_lfs_batch_info(
+            upload_infos=[op.upload_info for op in chunk_list],
+            repo_id=repo_id,
+            repo_type=repo_type,
+            revision=revision,
+            endpoint=endpoint,
+            headers=headers,
+            token=None,  # already passed in 'headers'
+            transfers=transfers,
+        )
+        if errors_chunk:
+            message = "\n".join(
+                [
+                    f"Encountered error for file with OID {err.get('oid')}: `{err.get('error', {}).get('message')}"
+                    for err in errors_chunk
+                ]
+            )
+            raise ValueError(f"LFS batch API returned errors:\n{message}")
+
+        # If server returns a transfer we didn't offer (e.g "xet" while uploading from BytesIO),
+        # fall back to LFS for this chunk.
+        if chosen_transfer == "xet" and ("xet" in transfers):
+            xet_additions.extend(chunk_list)
+        else:
+            lfs_actions.extend(actions_chunk)
+            for op in chunk_list:
+                lfs_oid2addop[op.upload_info.sha256.hex()] = op
+
+    if len(lfs_actions) > 0:
+        _upload_lfs_files(
+            actions=lfs_actions,
+            oid2addop=lfs_oid2addop,
+            headers=headers,
+            endpoint=endpoint,
+            num_threads=num_threads,
+        )
+
+    if len(xet_additions) > 0:
+        _upload_xet_files(
+            additions=xet_additions,
+            repo_type=repo_type,
+            repo_id=repo_id,
+            headers=headers,
+            endpoint=endpoint,
+            revision=revision,
+            create_pr=create_pr,
+        )
+
+
+@validate_hf_hub_args
+def _upload_lfs_files(
+    *,
+    actions: list[dict[str, Any]],
+    oid2addop: dict[str, CommitOperationAdd],
+    headers: dict[str, str],
+    endpoint: Optional[str] = None,
+    num_threads: int = 5,
 ):
     """
     Uploads the content of `additions` to the Hub using the large file storage protocol.
@@ -370,9 +451,21 @@ def _upload_lfs_files(
         - LFS Batch API: https://github.com/git-lfs/git-lfs/blob/main/docs/api/batch.md
 
     Args:
-        additions (`List` of `CommitOperationAdd`):
-            The files to be uploaded
-        repo_type (`str`):
+        actions (`list[dict[str, Any]]`):
+            LFS batch actions returned by the server.
+        oid2addop (`dict[str, CommitOperationAdd]`):
+            A dictionary mapping the OID of the file to the corresponding `CommitOperationAdd` object.
+        headers (`dict[str, str]`):
+            Headers to use for the request, including authorization headers and user agent.
+        endpoint (`str`, *optional*):
+            The endpoint to use for the request. Defaults to `constants.ENDPOINT`.
+        num_threads (`int`, *optional*):
+            The number of concurrent threads to use when uploading. Defaults to 5.
+
+    Raises:
+        [`EnvironmentError`](https://docs.python.org/3/library/exceptions.html#EnvironmentError)
+            If an upload failed for any reason
+        [`ValueError`](https://docs.python.org/3/library/exceptions.html#ValueError)
             Type of the repo to upload to: `"model"`, `"dataset"` or `"space"`.
         repo_id (`str`):
             A namespace (user or an organization) and a repo name separated
@@ -392,50 +485,17 @@ def _upload_lfs_files(
         [`HfHubHTTPError`]
             If the LFS batch endpoint returned an HTTP error.
     """
-    # Step 1: retrieve upload instructions from the LFS batch endpoint.
-    #         Upload instructions are retrieved by chunk of 256 files to avoid reaching
-    #         the payload limit.
-    batch_actions: list[dict] = []
-    for chunk in chunk_iterable(additions, chunk_size=UPLOAD_BATCH_MAX_NUM_FILES):
-        batch_actions_chunk, batch_errors_chunk = post_lfs_batch_info(
-            upload_infos=[op.upload_info for op in chunk],
-            repo_id=repo_id,
-            repo_type=repo_type,
-            revision=revision,
-            endpoint=endpoint,
-            headers=headers,
-            token=None,  # already passed in 'headers'
-        )
-
-        # If at least 1 error, we do not retrieve information for other chunks
-        if batch_errors_chunk:
-            message = "\n".join(
-                [
-                    f"Encountered error for file with OID {err.get('oid')}: `{err.get('error', {}).get('message')}"
-                    for err in batch_errors_chunk
-                ]
-            )
-            raise ValueError(f"LFS batch endpoint returned errors:\n{message}")
-
-        batch_actions += batch_actions_chunk
-    oid2addop = {add_op.upload_info.sha256.hex(): add_op for add_op in additions}
-
-    # Step 2: ignore files that have already been uploaded
+    # Filter out files already present upstream
     filtered_actions = []
-    for action in batch_actions:
+    for action in actions:
         if action.get("actions") is None:
             logger.debug(
-                f"Content of file {oid2addop[action['oid']].path_in_repo} is already"
-                " present upstream - skipping upload."
+                f"Content of file {oid2addop[action['oid']].path_in_repo} is already present upstream - skipping upload."
             )
         else:
             filtered_actions.append(action)
 
-    if len(filtered_actions) == 0:
-        logger.debug("No LFS files to upload.")
-        return
-
-    # Step 3: upload files concurrently according to these instructions
+    # Upload according to server-provided actions
     def _wrapped_lfs_upload(batch_action) -> None:
         try:
             operation = oid2addop[batch_action["oid"]]
@@ -479,7 +539,7 @@ def _upload_xet_files(
     This chunks the files and deduplicates the chunks before uploading them to xetcas storage.
 
     Args:
-        additions (`List` of `CommitOperationAdd`):
+        additions (`` of `CommitOperationAdd`):
             The files to be uploaded.
         repo_type (`str`):
             Type of the repo to upload to: `"model"`, `"dataset"` or `"space"`.
@@ -576,30 +636,30 @@ def _upload_xet_files(
         progress, progress_callback = None, None
 
     try:
-        for i, chunk in enumerate(chunk_iterable(additions, chunk_size=UPLOAD_BATCH_MAX_NUM_FILES)):
-            _chunk = [op for op in chunk]
-
-            bytes_ops = [op for op in _chunk if isinstance(op.path_or_fileobj, bytes)]
-            paths_ops = [op for op in _chunk if isinstance(op.path_or_fileobj, (str, Path))]
-
-            if len(paths_ops) > 0:
-                upload_files(
-                    [str(op.path_or_fileobj) for op in paths_ops],
-                    xet_endpoint,
-                    access_token_info,
-                    token_refresher,
-                    progress_callback,
-                    repo_type,
-                )
-            if len(bytes_ops) > 0:
-                upload_bytes(
-                    [op.path_or_fileobj for op in bytes_ops],
-                    xet_endpoint,
-                    access_token_info,
-                    token_refresher,
-                    progress_callback,
-                    repo_type,
-                )
+        all_bytes_ops = [op for op in additions if isinstance(op.path_or_fileobj, bytes)]
+        all_paths_ops = [op for op in additions if isinstance(op.path_or_fileobj, (str, Path))]
+
+        if len(all_paths_ops) > 0:
+            all_paths = [str(op.path_or_fileobj) for op in all_paths_ops]
+            upload_files(
+                all_paths,
+                xet_endpoint,
+                access_token_info,
+                token_refresher,
+                progress_callback,
+                repo_type,
+            )
+
+        if len(all_bytes_ops) > 0:
+            all_bytes = [op.path_or_fileobj for op in all_bytes_ops]
+            upload_bytes(
+                all_bytes,
+                xet_endpoint,
+                access_token_info,
+                token_refresher,
+                progress_callback,
+                repo_type,
+            )
 
     finally:
         if progress is not None:

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/src/huggingface_hub/_commit_scheduler.py

@@ -205,13 +205,10 @@ class CommitScheduler:
         """
         Push folder to the Hub and return the commit info.
 
-        <Tip warning={true}>
-
-        This method is not meant to be called directly. It is run in the background by the scheduler, respecting a
-        queue mechanism to avoid concurrent commits. Making a direct call to the method might lead to concurrency
-        issues.
-
-        </Tip>
+        > [!WARNING]
+        > This method is not meant to be called directly. It is run in the background by the scheduler, respecting a
+        > queue mechanism to avoid concurrent commits. Making a direct call to the method might lead to concurrency
+        > issues.
 
         The default behavior of `push_to_hub` is to assume an append-only folder. It lists all files in the folder and
         uploads only changed files. If no changes are found, the method returns without committing anything. If you want

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/src/huggingface_hub/_login.py

@@ -20,7 +20,7 @@ from pathlib import Path
 from typing import Optional
 
 from . import constants
-from .commands._cli_utils import ANSI
+from .cli._cli_utils import ANSI
 from .utils import (
     capture_output,
     get_token,
@@ -70,21 +70,15 @@ def login(
     To log in from outside of a script, one can also use `hf auth login` which is
     a cli command that wraps [`login`].
 
-    <Tip>
+    > [!TIP]
+    > [`login`] is a drop-in replacement method for [`notebook_login`] as it wraps and
+    > extends its capabilities.
 
-    [`login`] is a drop-in replacement method for [`notebook_login`] as it wraps and
-    extends its capabilities.
-
-    </Tip>
-
-    <Tip>
-
-    When the token is not passed, [`login`] will automatically detect if the script runs
-    in a notebook or not. However, this detection might not be accurate due to the
-    variety of notebooks that exists nowadays. If that is the case, you can always force
-    the UI by using [`notebook_login`] or [`interpreter_login`].
-
-    </Tip>
+    > [!TIP]
+    > When the token is not passed, [`login`] will automatically detect if the script runs
+    > in a notebook or not. However, this detection might not be accurate due to the
+    > variety of notebooks that exists nowadays. If that is the case, you can always force
+    > the UI by using [`notebook_login`] or [`interpreter_login`].
 
     Args:
         token (`str`, *optional*):
@@ -250,7 +244,7 @@ def interpreter_login(*, skip_if_logged_in: bool = False) -> None:
         logger.info("User is already logged in.")
         return
 
-    from .commands.delete_cache import _ask_for_confirmation_no_tui
+    from .cli.cache import _ask_for_confirmation_no_tui
 
     print(_HF_LOGO_ASCII)
     if get_token() is not None:

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/src/huggingface_hub/_snapshot_download.py

@@ -1,6 +1,6 @@
 import os
 from pathlib import Path
-from typing import Iterable, Optional, Union
+from typing import Iterable, List, Literal, Optional, Union, overload
 
 import httpx
 from tqdm.auto import tqdm as base_tqdm
@@ -8,13 +8,14 @@ from tqdm.contrib.concurrent import thread_map
 
 from . import constants
 from .errors import (
+    DryRunError,
     GatedRepoError,
     HfHubHTTPError,
     LocalEntryNotFoundError,
     RepositoryNotFoundError,
     RevisionNotFoundError,
 )
-from .file_download import REGEX_COMMIT_HASH, hf_hub_download, repo_folder_name
+from .file_download import REGEX_COMMIT_HASH, DryRunFileInfo, hf_hub_download, repo_folder_name
 from .hf_api import DatasetInfo, HfApi, ModelInfo, RepoFile, SpaceInfo
 from .utils import OfflineModeIsEnabled, filter_repo_objects, logging, validate_hf_hub_args
 from .utils import tqdm as hf_tqdm
@@ -25,6 +26,81 @@ logger = logging.get_logger(__name__)
 VERY_LARGE_REPO_THRESHOLD = 50000  # After this limit, we don't consider `repo_info.siblings` to be reliable enough
 
 
+@overload
+def snapshot_download(
+    repo_id: str,
+    *,
+    repo_type: Optional[str] = None,
+    revision: Optional[str] = None,
+    cache_dir: Union[str, Path, None] = None,
+    local_dir: Union[str, Path, None] = None,
+    library_name: Optional[str] = None,
+    library_version: Optional[str] = None,
+    user_agent: Optional[Union[dict, str]] = None,
+    etag_timeout: float = constants.DEFAULT_ETAG_TIMEOUT,
+    force_download: bool = False,
+    token: Optional[Union[bool, str]] = None,
+    local_files_only: bool = False,
+    allow_patterns: Optional[Union[list[str], str]] = None,
+    ignore_patterns: Optional[Union[list[str], str]] = None,
+    max_workers: int = 8,
+    tqdm_class: Optional[type[base_tqdm]] = None,
+    headers: Optional[dict[str, str]] = None,
+    endpoint: Optional[str] = None,
+    dry_run: Literal[False] = False,
+) -> str: ...
+
+
+@overload
+def snapshot_download(
+    repo_id: str,
+    *,
+    repo_type: Optional[str] = None,
+    revision: Optional[str] = None,
+    cache_dir: Union[str, Path, None] = None,
+    local_dir: Union[str, Path, None] = None,
+    library_name: Optional[str] = None,
+    library_version: Optional[str] = None,
+    user_agent: Optional[Union[dict, str]] = None,
+    etag_timeout: float = constants.DEFAULT_ETAG_TIMEOUT,
+    force_download: bool = False,
+    token: Optional[Union[bool, str]] = None,
+    local_files_only: bool = False,
+    allow_patterns: Optional[Union[list[str], str]] = None,
+    ignore_patterns: Optional[Union[list[str], str]] = None,
+    max_workers: int = 8,
+    tqdm_class: Optional[type[base_tqdm]] = None,
+    headers: Optional[dict[str, str]] = None,
+    endpoint: Optional[str] = None,
+    dry_run: Literal[True] = True,
+) -> list[DryRunFileInfo]: ...
+
+
+@overload
+def snapshot_download(
+    repo_id: str,
+    *,
+    repo_type: Optional[str] = None,
+    revision: Optional[str] = None,
+    cache_dir: Union[str, Path, None] = None,
+    local_dir: Union[str, Path, None] = None,
+    library_name: Optional[str] = None,
+    library_version: Optional[str] = None,
+    user_agent: Optional[Union[dict, str]] = None,
+    etag_timeout: float = constants.DEFAULT_ETAG_TIMEOUT,
+    force_download: bool = False,
+    token: Optional[Union[bool, str]] = None,
+    local_files_only: bool = False,
+    allow_patterns: Optional[Union[list[str], str]] = None,
+    ignore_patterns: Optional[Union[list[str], str]] = None,
+    max_workers: int = 8,
+    tqdm_class: Optional[type[base_tqdm]] = None,
+    headers: Optional[dict[str, str]] = None,
+    endpoint: Optional[str] = None,
+    dry_run: bool = False,
+) -> Union[str, list[DryRunFileInfo]]: ...
+
+
 @validate_hf_hub_args
 def snapshot_download(
     repo_id: str,
@@ -46,7 +122,8 @@ def snapshot_download(
     tqdm_class: Optional[type[base_tqdm]] = None,
     headers: Optional[dict[str, str]] = None,
     endpoint: Optional[str] = None,
-) -> str:
+    dry_run: bool = False,
+) -> Union[str, list[DryRunFileInfo]]:
     """Download repo files.
 
     Download a whole snapshot of a repo's files at the specified revision. This is useful when you want all files from
@@ -109,9 +186,14 @@ def snapshot_download(
             Note that the `tqdm_class` is not passed to each individual download.
             Defaults to the custom HF progress bar that can be disabled by setting
             `HF_HUB_DISABLE_PROGRESS_BARS` environment variable.
+        dry_run (`bool`, *optional*, defaults to `False`):
+            If `True`, perform a dry run without actually downloading the files. Returns a list of
+            [`DryRunFileInfo`] objects containing information about what would be downloaded.
 
     Returns:
-        `str`: folder path of the repo snapshot.
+        `str` or list of [`DryRunFileInfo`]:
+            - If `dry_run=False`: Local snapshot path.
+            - If `dry_run=True`: A list of [`DryRunFileInfo`] objects containing download information.
 
     Raises:
         [`~utils.RepositoryNotFoundError`]
@@ -187,6 +269,11 @@ def snapshot_download(
     #    - f the specified revision is a branch or tag, look inside "refs".
     # => if local_dir is not None, we will return the path to the local folder if it exists.
     if repo_info is None:
+        if dry_run:
+            raise DryRunError(
+                "Dry run cannot be performed as the repository cannot be accessed. Please check your internet connection or authentication token."
+            ) from api_call_error
+
         # Try to get which commit hash corresponds to the specified revision
         commit_hash = None
         if REGEX_COMMIT_HASH.match(revision):
@@ -273,6 +360,8 @@ def snapshot_download(
         tqdm_desc = f"Fetching {len(filtered_repo_files)} files"
     else:
         tqdm_desc = "Fetching ... files"
+    if dry_run:
+        tqdm_desc = "[dry-run] " + tqdm_desc
 
     commit_hash = repo_info.sha
     snapshot_folder = os.path.join(storage_folder, "snapshots", commit_hash)
@@ -288,28 +377,33 @@ def snapshot_download(
         except OSError as e:
             logger.warning(f"Ignored error while writing commit hash to {ref_path}: {e}.")
 
+    results: List[Union[str, DryRunFileInfo]] = []
+
     # we pass the commit_hash to hf_hub_download
     # so no network call happens if we already
     # have the file locally.
-    def _inner_hf_hub_download(repo_file: str):
-        return hf_hub_download(
-            repo_id,
-            filename=repo_file,
-            repo_type=repo_type,
-            revision=commit_hash,
-            endpoint=endpoint,
-            cache_dir=cache_dir,
-            local_dir=local_dir,
-            library_name=library_name,
-            library_version=library_version,
-            user_agent=user_agent,
-            etag_timeout=etag_timeout,
-            force_download=force_download,
-            token=token,
-            headers=headers,
+    def _inner_hf_hub_download(repo_file: str) -> None:
+        results.append(
+            hf_hub_download(  # type: ignore[no-matching-overload] # ty not happy, don't know why :/
+                repo_id,
+                filename=repo_file,
+                repo_type=repo_type,
+                revision=commit_hash,
+                endpoint=endpoint,
+                cache_dir=cache_dir,
+                local_dir=local_dir,
+                library_name=library_name,
+                library_version=library_version,
+                user_agent=user_agent,
+                etag_timeout=etag_timeout,
+                force_download=force_download,
+                token=token,
+                headers=headers,
+                dry_run=dry_run,
+            )
         )
 
-    if constants.HF_HUB_ENABLE_HF_TRANSFER:
+    if constants.HF_HUB_ENABLE_HF_TRANSFER and not dry_run:
         # when using hf_transfer we don't want extra parallelism
         # from the one hf_transfer provides
         for file in filtered_repo_files:
@@ -324,6 +418,10 @@ def snapshot_download(
             tqdm_class=tqdm_class or hf_tqdm,
         )
 
+    if dry_run:
+        assert all(isinstance(r, DryRunFileInfo) for r in results)
+        return results  # type: ignore
+
     if local_dir is not None:
         return str(os.path.realpath(local_dir))
     return snapshot_folder

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/src/huggingface_hub/_tensorboard_logger.py

@@ -52,11 +52,8 @@ class HFSummaryWriter(_RuntimeSummaryWriter):
     issue), the main script will not be interrupted. Data is automatically pushed to the Hub every `commit_every`
     minutes (default to every 5 minutes).
 
-    <Tip warning={true}>
-
-    `HFSummaryWriter` is experimental. Its API is subject to change in the future without prior notice.
-
-    </Tip>
+    > [!WARNING]
+    > `HFSummaryWriter` is experimental. Its API is subject to change in the future without prior notice.
 
     Args:
         repo_id (`str`):

{huggingface_hub-1.0.0rc1 → huggingface_hub-1.0.0rc3}/src/huggingface_hub/_upload_large_folder.py

@@ -31,8 +31,7 @@ from . import constants
 from ._commit_api import CommitOperationAdd, UploadInfo, _fetch_upload_modes
 from ._local_folder import LocalUploadFileMetadata, LocalUploadFilePaths, get_local_upload_paths, read_upload_metadata
 from .constants import DEFAULT_REVISION, REPO_TYPES
-from .utils import DEFAULT_IGNORE_PATTERNS, filter_repo_objects, tqdm
-from .utils._cache_manager import _format_size
+from .utils import DEFAULT_IGNORE_PATTERNS, _format_size, filter_repo_objects, tqdm
 from .utils._runtime import is_xet_available
 from .utils.sha import sha_fileobj