chore: 添加虚拟环境到仓库

- 添加 backend_service/venv 虚拟环境
- 包含所有Python依赖包
- 注意：虚拟环境约393MB，包含12655个文件

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/arrow.py

@@ -0,0 +1,307 @@
+import errno
+import io
+import os
+import secrets
+import shutil
+from contextlib import suppress
+from functools import cached_property, wraps
+from urllib.parse import parse_qs
+
+from fsspec.spec import AbstractFileSystem
+from fsspec.utils import (
+    get_package_version_without_import,
+    infer_storage_options,
+    mirror_from,
+    tokenize,
+)
+
+
+def wrap_exceptions(func):
+    @wraps(func)
+    def wrapper(*args, **kwargs):
+        try:
+            return func(*args, **kwargs)
+        except OSError as exception:
+            if not exception.args:
+                raise
+
+            message, *args = exception.args
+            if isinstance(message, str) and "does not exist" in message:
+                raise FileNotFoundError(errno.ENOENT, message) from exception
+            else:
+                raise
+
+    return wrapper
+
+
+PYARROW_VERSION = None
+
+
+class ArrowFSWrapper(AbstractFileSystem):
+    """FSSpec-compatible wrapper of pyarrow.fs.FileSystem.
+
+    Parameters
+    ----------
+    fs : pyarrow.fs.FileSystem
+
+    """
+
+    root_marker = "/"
+
+    def __init__(self, fs, **kwargs):
+        global PYARROW_VERSION
+        PYARROW_VERSION = get_package_version_without_import("pyarrow")
+        self.fs = fs
+        super().__init__(**kwargs)
+
+    @property
+    def protocol(self):
+        return self.fs.type_name
+
+    @cached_property
+    def fsid(self):
+        return "hdfs_" + tokenize(self.fs.host, self.fs.port)
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        ops = infer_storage_options(path)
+        path = ops["path"]
+        if path.startswith("//"):
+            # special case for "hdfs://path" (without the triple slash)
+            path = path[1:]
+        return path
+
+    def ls(self, path, detail=False, **kwargs):
+        path = self._strip_protocol(path)
+        from pyarrow.fs import FileSelector
+
+        try:
+            entries = [
+                self._make_entry(entry)
+                for entry in self.fs.get_file_info(FileSelector(path))
+            ]
+        except (FileNotFoundError, NotADirectoryError):
+            entries = [self.info(path, **kwargs)]
+        if detail:
+            return entries
+        else:
+            return [entry["name"] for entry in entries]
+
+    def info(self, path, **kwargs):
+        path = self._strip_protocol(path)
+        [info] = self.fs.get_file_info([path])
+        return self._make_entry(info)
+
+    def exists(self, path):
+        path = self._strip_protocol(path)
+        try:
+            self.info(path)
+        except FileNotFoundError:
+            return False
+        else:
+            return True
+
+    def _make_entry(self, info):
+        from pyarrow.fs import FileType
+
+        if info.type is FileType.Directory:
+            kind = "directory"
+        elif info.type is FileType.File:
+            kind = "file"
+        elif info.type is FileType.NotFound:
+            raise FileNotFoundError(errno.ENOENT, os.strerror(errno.ENOENT), info.path)
+        else:
+            kind = "other"
+
+        return {
+            "name": info.path,
+            "size": info.size,
+            "type": kind,
+            "mtime": info.mtime,
+        }
+
+    @wrap_exceptions
+    def cp_file(self, path1, path2, **kwargs):
+        path1 = self._strip_protocol(path1).rstrip("/")
+        path2 = self._strip_protocol(path2).rstrip("/")
+
+        with self._open(path1, "rb") as lstream:
+            tmp_fname = f"{path2}.tmp.{secrets.token_hex(6)}"
+            try:
+                with self.open(tmp_fname, "wb") as rstream:
+                    shutil.copyfileobj(lstream, rstream)
+                self.fs.move(tmp_fname, path2)
+            except BaseException:
+                with suppress(FileNotFoundError):
+                    self.fs.delete_file(tmp_fname)
+                raise
+
+    @wrap_exceptions
+    def mv(self, path1, path2, **kwargs):
+        path1 = self._strip_protocol(path1).rstrip("/")
+        path2 = self._strip_protocol(path2).rstrip("/")
+        self.fs.move(path1, path2)
+
+    @wrap_exceptions
+    def rm_file(self, path):
+        path = self._strip_protocol(path)
+        self.fs.delete_file(path)
+
+    @wrap_exceptions
+    def rm(self, path, recursive=False, maxdepth=None):
+        path = self._strip_protocol(path).rstrip("/")
+        if self.isdir(path):
+            if recursive:
+                self.fs.delete_dir(path)
+            else:
+                raise ValueError("Can't delete directories without recursive=False")
+        else:
+            self.fs.delete_file(path)
+
+    @wrap_exceptions
+    def _open(self, path, mode="rb", block_size=None, seekable=True, **kwargs):
+        if mode == "rb":
+            if seekable:
+                method = self.fs.open_input_file
+            else:
+                method = self.fs.open_input_stream
+        elif mode == "wb":
+            method = self.fs.open_output_stream
+        elif mode == "ab":
+            method = self.fs.open_append_stream
+        else:
+            raise ValueError(f"unsupported mode for Arrow filesystem: {mode!r}")
+
+        _kwargs = {}
+        if mode != "rb" or not seekable:
+            if int(PYARROW_VERSION.split(".")[0]) >= 4:
+                # disable compression auto-detection
+                _kwargs["compression"] = None
+        stream = method(path, **_kwargs)
+
+        return ArrowFile(self, stream, path, mode, block_size, **kwargs)
+
+    @wrap_exceptions
+    def mkdir(self, path, create_parents=True, **kwargs):
+        path = self._strip_protocol(path)
+        if create_parents:
+            self.makedirs(path, exist_ok=True)
+        else:
+            self.fs.create_dir(path, recursive=False)
+
+    @wrap_exceptions
+    def makedirs(self, path, exist_ok=False):
+        path = self._strip_protocol(path)
+        self.fs.create_dir(path, recursive=True)
+
+    @wrap_exceptions
+    def rmdir(self, path):
+        path = self._strip_protocol(path)
+        self.fs.delete_dir(path)
+
+    @wrap_exceptions
+    def modified(self, path):
+        path = self._strip_protocol(path)
+        return self.fs.get_file_info(path).mtime
+
+    def cat_file(self, path, start=None, end=None, **kwargs):
+        kwargs["seekable"] = start not in [None, 0]
+        return super().cat_file(path, start=None, end=None, **kwargs)
+
+    def get_file(self, rpath, lpath, **kwargs):
+        kwargs["seekable"] = False
+        super().get_file(rpath, lpath, **kwargs)
+
+
+@mirror_from(
+    "stream",
+    [
+        "read",
+        "seek",
+        "tell",
+        "write",
+        "readable",
+        "writable",
+        "close",
+        "size",
+        "seekable",
+    ],
+)
+class ArrowFile(io.IOBase):
+    def __init__(self, fs, stream, path, mode, block_size=None, **kwargs):
+        self.path = path
+        self.mode = mode
+
+        self.fs = fs
+        self.stream = stream
+
+        self.blocksize = self.block_size = block_size
+        self.kwargs = kwargs
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, *args):
+        return self.close()
+
+
+class HadoopFileSystem(ArrowFSWrapper):
+    """A wrapper on top of the pyarrow.fs.HadoopFileSystem
+    to connect it's interface with fsspec"""
+
+    protocol = "hdfs"
+
+    def __init__(
+        self,
+        host="default",
+        port=0,
+        user=None,
+        kerb_ticket=None,
+        replication=3,
+        extra_conf=None,
+        **kwargs,
+    ):
+        """
+
+        Parameters
+        ----------
+        host: str
+            Hostname, IP or "default" to try to read from Hadoop config
+        port: int
+            Port to connect on, or default from Hadoop config if 0
+        user: str or None
+            If given, connect as this username
+        kerb_ticket: str or None
+            If given, use this ticket for authentication
+        replication: int
+            set replication factor of file for write operations. default value is 3.
+        extra_conf: None or dict
+            Passed on to HadoopFileSystem
+        """
+        from pyarrow.fs import HadoopFileSystem
+
+        fs = HadoopFileSystem(
+            host=host,
+            port=port,
+            user=user,
+            kerb_ticket=kerb_ticket,
+            replication=replication,
+            extra_conf=extra_conf,
+        )
+        super().__init__(fs=fs, **kwargs)
+
+    @staticmethod
+    def _get_kwargs_from_urls(path):
+        ops = infer_storage_options(path)
+        out = {}
+        if ops.get("host", None):
+            out["host"] = ops["host"]
+        if ops.get("username", None):
+            out["user"] = ops["username"]
+        if ops.get("port", None):
+            out["port"] = ops["port"]
+        if ops.get("url_query", None):
+            queries = parse_qs(ops["url_query"])
+            if queries.get("replication", None):
+                out["replication"] = int(queries["replication"][0])
+        return out

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/asyn_wrapper.py

@@ -0,0 +1,122 @@
+import asyncio
+import functools
+import inspect
+
+import fsspec
+from fsspec.asyn import AsyncFileSystem, running_async
+
+
+def async_wrapper(func, obj=None, semaphore=None):
+    """
+    Wraps a synchronous function to make it awaitable.
+
+    Parameters
+    ----------
+    func : callable
+        The synchronous function to wrap.
+    obj : object, optional
+        The instance to bind the function to, if applicable.
+    semaphore : asyncio.Semaphore, optional
+        A semaphore to limit concurrent calls.
+
+    Returns
+    -------
+    coroutine
+        An awaitable version of the function.
+    """
+
+    @functools.wraps(func)
+    async def wrapper(*args, **kwargs):
+        if semaphore:
+            async with semaphore:
+                return await asyncio.to_thread(func, *args, **kwargs)
+        return await asyncio.to_thread(func, *args, **kwargs)
+
+    return wrapper
+
+
+class AsyncFileSystemWrapper(AsyncFileSystem):
+    """
+    A wrapper class to convert a synchronous filesystem into an asynchronous one.
+
+    This class takes an existing synchronous filesystem implementation and wraps all
+    its methods to provide an asynchronous interface.
+
+    Parameters
+    ----------
+    sync_fs : AbstractFileSystem
+        The synchronous filesystem instance to wrap.
+    """
+
+    protocol = "asyncwrapper", "async_wrapper"
+    cachable = False
+
+    def __init__(
+        self,
+        fs=None,
+        asynchronous=None,
+        target_protocol=None,
+        target_options=None,
+        semaphore=None,
+        max_concurrent_tasks=None,
+        **kwargs,
+    ):
+        if asynchronous is None:
+            asynchronous = running_async()
+        super().__init__(asynchronous=asynchronous, **kwargs)
+        if fs is not None:
+            self.sync_fs = fs
+        else:
+            self.sync_fs = fsspec.filesystem(target_protocol, **target_options)
+        self.protocol = self.sync_fs.protocol
+        self.semaphore = semaphore
+        self._wrap_all_sync_methods()
+
+    @property
+    def fsid(self):
+        return f"async_{self.sync_fs.fsid}"
+
+    def _wrap_all_sync_methods(self):
+        """
+        Wrap all synchronous methods of the underlying filesystem with asynchronous versions.
+        """
+        excluded_methods = {"open"}
+        for method_name in dir(self.sync_fs):
+            if method_name.startswith("_") or method_name in excluded_methods:
+                continue
+
+            attr = inspect.getattr_static(self.sync_fs, method_name)
+            if isinstance(attr, property):
+                continue
+
+            method = getattr(self.sync_fs, method_name)
+            if callable(method) and not inspect.iscoroutinefunction(method):
+                async_method = async_wrapper(method, obj=self, semaphore=self.semaphore)
+                setattr(self, f"_{method_name}", async_method)
+
+    @classmethod
+    def wrap_class(cls, sync_fs_class):
+        """
+        Create a new class that can be used to instantiate an AsyncFileSystemWrapper
+        with lazy instantiation of the underlying synchronous filesystem.
+
+        Parameters
+        ----------
+        sync_fs_class : type
+            The class of the synchronous filesystem to wrap.
+
+        Returns
+        -------
+        type
+            A new class that wraps the provided synchronous filesystem class.
+        """
+
+        class GeneratedAsyncFileSystemWrapper(cls):
+            def __init__(self, *args, **kwargs):
+                sync_fs = sync_fs_class(*args, **kwargs)
+                super().__init__(sync_fs)
+
+        GeneratedAsyncFileSystemWrapper.__name__ = (
+            f"Async{sync_fs_class.__name__}Wrapper"
+        )
+        return GeneratedAsyncFileSystemWrapper

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/cache_mapper.py

@@ -0,0 +1,75 @@
+from __future__ import annotations
+
+import abc
+import hashlib
+
+from fsspec.implementations.local import make_path_posix
+
+
+class AbstractCacheMapper(abc.ABC):
+    """Abstract super-class for mappers from remote URLs to local cached
+    basenames.
+    """
+
+    @abc.abstractmethod
+    def __call__(self, path: str) -> str: ...
+
+    def __eq__(self, other: object) -> bool:
+        # Identity only depends on class. When derived classes have attributes
+        # they will need to be included.
+        return isinstance(other, type(self))
+
+    def __hash__(self) -> int:
+        # Identity only depends on class. When derived classes have attributes
+        # they will need to be included.
+        return hash(type(self))
+
+
+class BasenameCacheMapper(AbstractCacheMapper):
+    """Cache mapper that uses the basename of the remote URL and a fixed number
+    of directory levels above this.
+
+    The default is zero directory levels, meaning different paths with the same
+    basename will have the same cached basename.
+    """
+
+    def __init__(self, directory_levels: int = 0):
+        if directory_levels < 0:
+            raise ValueError(
+                "BasenameCacheMapper requires zero or positive directory_levels"
+            )
+        self.directory_levels = directory_levels
+
+        # Separator for directories when encoded as strings.
+        self._separator = "_@_"
+
+    def __call__(self, path: str) -> str:
+        path = make_path_posix(path)
+        prefix, *bits = path.rsplit("/", self.directory_levels + 1)
+        if bits:
+            return self._separator.join(bits)
+        else:
+            return prefix  # No separator found, simple filename
+
+    def __eq__(self, other: object) -> bool:
+        return super().__eq__(other) and self.directory_levels == other.directory_levels
+
+    def __hash__(self) -> int:
+        return super().__hash__() ^ hash(self.directory_levels)
+
+
+class HashCacheMapper(AbstractCacheMapper):
+    """Cache mapper that uses a hash of the remote URL."""
+
+    def __call__(self, path: str) -> str:
+        return hashlib.sha256(path.encode()).hexdigest()
+
+
+def create_cache_mapper(same_names: bool) -> AbstractCacheMapper:
+    """Factory method to create cache mapper for backward compatibility with
+    ``CachingFileSystem`` constructor using ``same_names`` kwarg.
+    """
+    if same_names:
+        return BasenameCacheMapper()
+    else:
+        return HashCacheMapper()

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/cache_metadata.py

@@ -0,0 +1,233 @@
+from __future__ import annotations
+
+import os
+import pickle
+import time
+from typing import TYPE_CHECKING
+
+from fsspec.utils import atomic_write
+
+try:
+    import ujson as json
+except ImportError:
+    if not TYPE_CHECKING:
+        import json
+
+if TYPE_CHECKING:
+    from collections.abc import Iterator
+    from typing import Any, Literal
+
+    from typing_extensions import TypeAlias
+
+    from .cached import CachingFileSystem
+
+    Detail: TypeAlias = dict[str, Any]
+
+
+class CacheMetadata:
+    """Cache metadata.
+
+    All reading and writing of cache metadata is performed by this class,
+    accessing the cached files and blocks is not.
+
+    Metadata is stored in a single file per storage directory in JSON format.
+    For backward compatibility, also reads metadata stored in pickle format
+    which is converted to JSON when next saved.
+    """
+
+    def __init__(self, storage: list[str]):
+        """
+
+        Parameters
+        ----------
+        storage: list[str]
+            Directories containing cached files, must be at least one. Metadata
+            is stored in the last of these directories by convention.
+        """
+        if not storage:
+            raise ValueError("CacheMetadata expects at least one storage location")
+
+        self._storage = storage
+        self.cached_files: list[Detail] = [{}]
+
+        # Private attribute to force saving of metadata in pickle format rather than
+        # JSON for use in tests to confirm can read both pickle and JSON formats.
+        self._force_save_pickle = False
+
+    def _load(self, fn: str) -> Detail:
+        """Low-level function to load metadata from specific file"""
+        try:
+            with open(fn, "r") as f:
+                loaded = json.load(f)
+        except ValueError:
+            with open(fn, "rb") as f:
+                loaded = pickle.load(f)
+        for c in loaded.values():
+            if isinstance(c.get("blocks"), list):
+                c["blocks"] = set(c["blocks"])
+        return loaded
+
+    def _save(self, metadata_to_save: Detail, fn: str) -> None:
+        """Low-level function to save metadata to specific file"""
+        if self._force_save_pickle:
+            with atomic_write(fn) as f:
+                pickle.dump(metadata_to_save, f)
+        else:
+            with atomic_write(fn, mode="w") as f:
+                json.dump(metadata_to_save, f)
+
+    def _scan_locations(
+        self, writable_only: bool = False
+    ) -> Iterator[tuple[str, str, bool]]:
+        """Yield locations (filenames) where metadata is stored, and whether
+        writable or not.
+
+        Parameters
+        ----------
+        writable: bool
+            Set to True to only yield writable locations.
+
+        Returns
+        -------
+        Yields (str, str, bool)
+        """
+        n = len(self._storage)
+        for i, storage in enumerate(self._storage):
+            writable = i == n - 1
+            if writable_only and not writable:
+                continue
+            yield os.path.join(storage, "cache"), storage, writable
+
+    def check_file(
+        self, path: str, cfs: CachingFileSystem | None
+    ) -> Literal[False] | tuple[Detail, str]:
+        """If path is in cache return its details, otherwise return ``False``.
+
+        If the optional CachingFileSystem is specified then it is used to
+        perform extra checks to reject possible matches, such as if they are
+        too old.
+        """
+        for (fn, base, _), cache in zip(self._scan_locations(), self.cached_files):
+            if path not in cache:
+                continue
+            detail = cache[path].copy()
+
+            if cfs is not None:
+                if cfs.check_files and detail["uid"] != cfs.fs.ukey(path):
+                    # Wrong file as determined by hash of file properties
+                    continue
+                if cfs.expiry and time.time() - detail["time"] > cfs.expiry:
+                    # Cached file has expired
+                    continue
+
+            fn = os.path.join(base, detail["fn"])
+            if os.path.exists(fn):
+                return detail, fn
+        return False
+
+    def clear_expired(self, expiry_time: int) -> tuple[list[str], bool]:
+        """Remove expired metadata from the cache.
+
+        Returns names of files corresponding to expired metadata and a boolean
+        flag indicating whether the writable cache is empty. Caller is
+        responsible for deleting the expired files.
+        """
+        expired_files = []
+        for path, detail in self.cached_files[-1].copy().items():
+            if time.time() - detail["time"] > expiry_time:
+                fn = detail.get("fn", "")
+                if not fn:
+                    raise RuntimeError(
+                        f"Cache metadata does not contain 'fn' for {path}"
+                    )
+                fn = os.path.join(self._storage[-1], fn)
+                expired_files.append(fn)
+                self.cached_files[-1].pop(path)
+
+        if self.cached_files[-1]:
+            cache_path = os.path.join(self._storage[-1], "cache")
+            self._save(self.cached_files[-1], cache_path)
+
+        writable_cache_empty = not self.cached_files[-1]
+        return expired_files, writable_cache_empty
+
+    def load(self) -> None:
+        """Load all metadata from disk and store in ``self.cached_files``"""
+        cached_files = []
+        for fn, _, _ in self._scan_locations():
+            if os.path.exists(fn):
+                # TODO: consolidate blocks here
+                cached_files.append(self._load(fn))
+            else:
+                cached_files.append({})
+        self.cached_files = cached_files or [{}]
+
+    def on_close_cached_file(self, f: Any, path: str) -> None:
+        """Perform side-effect actions on closing a cached file.
+
+        The actual closing of the file is the responsibility of the caller.
+        """
+        # File must be writeble, so in self.cached_files[-1]
+        c = self.cached_files[-1][path]
+        if c["blocks"] is not True and len(c["blocks"]) * f.blocksize >= f.size:
+            c["blocks"] = True
+
+    def pop_file(self, path: str) -> str | None:
+        """Remove metadata of cached file.
+
+        If path is in the cache, return the filename of the cached file,
+        otherwise return ``None``.  Caller is responsible for deleting the
+        cached file.
+        """
+        details = self.check_file(path, None)
+        if not details:
+            return None
+        _, fn = details
+        if fn.startswith(self._storage[-1]):
+            self.cached_files[-1].pop(path)
+            self.save()
+        else:
+            raise PermissionError(
+                "Can only delete cached file in last, writable cache location"
+            )
+        return fn
+
+    def save(self) -> None:
+        """Save metadata to disk"""
+        for (fn, _, writable), cache in zip(self._scan_locations(), self.cached_files):
+            if not writable:
+                continue
+
+            if os.path.exists(fn):
+                cached_files = self._load(fn)
+                for k, c in cached_files.items():
+                    if k in cache:
+                        if c["blocks"] is True or cache[k]["blocks"] is True:
+                            c["blocks"] = True
+                        else:
+                            # self.cached_files[*][*]["blocks"] must continue to
+                            # point to the same set object so that updates
+                            # performed by MMapCache are propagated back to
+                            # self.cached_files.
+                            blocks = cache[k]["blocks"]
+                            blocks.update(c["blocks"])
+                            c["blocks"] = blocks
+                        c["time"] = max(c["time"], cache[k]["time"])
+                        c["uid"] = cache[k]["uid"]
+
+                # Files can be added to cache after it was written once
+                for k, c in cache.items():
+                    if k not in cached_files:
+                        cached_files[k] = c
+            else:
+                cached_files = cache
+            cache = {k: v.copy() for k, v in cached_files.items()}
+            for c in cache.values():
+                if isinstance(c["blocks"], set):
+                    c["blocks"] = list(c["blocks"])
+            self._save(cache, fn)
+            self.cached_files[-1] = cached_files
+
+    def update_file(self, path: str, detail: Detail) -> None:
+        """Update metadata for specific file in memory, do not save"""
+        self.cached_files[-1][path] = detail

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/chained.py

@@ -0,0 +1,23 @@
+from typing import ClassVar
+
+from fsspec import AbstractFileSystem
+
+__all__ = ("ChainedFileSystem",)
+
+
+class ChainedFileSystem(AbstractFileSystem):
+    """Chained filesystem base class.
+
+    A chained filesystem is designed to be layered over another FS.
+    This is useful to implement things like caching.
+
+    This base class does very little on its own, but is used as a marker
+    that the class is designed for chaining.
+
+    Right now this is only used in `url_to_fs` to provide the path argument
+    (`fo`) to the chained filesystem from the underlying filesystem.
+
+    Additional functionality may be added in the future.
+    """
+
+    protocol: ClassVar[str] = "chained"

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/dask.py

@@ -0,0 +1,152 @@
+import dask
+from distributed.client import Client, _get_global_client
+from distributed.worker import Worker
+
+from fsspec import filesystem
+from fsspec.spec import AbstractBufferedFile, AbstractFileSystem
+from fsspec.utils import infer_storage_options
+
+
+def _get_client(client):
+    if client is None:
+        return _get_global_client()
+    elif isinstance(client, Client):
+        return client
+    else:
+        # e.g., connection string
+        return Client(client)
+
+
+def _in_worker():
+    return bool(Worker._instances)
+
+
+class DaskWorkerFileSystem(AbstractFileSystem):
+    """View files accessible to a worker as any other remote file-system
+
+    When instances are run on the worker, uses the real filesystem. When
+    run on the client, they call the worker to provide information or data.
+
+    **Warning** this implementation is experimental, and read-only for now.
+    """
+
+    def __init__(
+        self, target_protocol=None, target_options=None, fs=None, client=None, **kwargs
+    ):
+        super().__init__(**kwargs)
+        if not (fs is None) ^ (target_protocol is None):
+            raise ValueError(
+                "Please provide one of filesystem instance (fs) or"
+                " target_protocol, not both"
+            )
+        self.target_protocol = target_protocol
+        self.target_options = target_options
+        self.worker = None
+        self.client = client
+        self.fs = fs
+        self._determine_worker()
+
+    @staticmethod
+    def _get_kwargs_from_urls(path):
+        so = infer_storage_options(path)
+        if "host" in so and "port" in so:
+            return {"client": f"{so['host']}:{so['port']}"}
+        else:
+            return {}
+
+    def _determine_worker(self):
+        if _in_worker():
+            self.worker = True
+            if self.fs is None:
+                self.fs = filesystem(
+                    self.target_protocol, **(self.target_options or {})
+                )
+        else:
+            self.worker = False
+            self.client = _get_client(self.client)
+            self.rfs = dask.delayed(self)
+
+    def mkdir(self, *args, **kwargs):
+        if self.worker:
+            self.fs.mkdir(*args, **kwargs)
+        else:
+            self.rfs.mkdir(*args, **kwargs).compute()
+
+    def rm(self, *args, **kwargs):
+        if self.worker:
+            self.fs.rm(*args, **kwargs)
+        else:
+            self.rfs.rm(*args, **kwargs).compute()
+
+    def copy(self, *args, **kwargs):
+        if self.worker:
+            self.fs.copy(*args, **kwargs)
+        else:
+            self.rfs.copy(*args, **kwargs).compute()
+
+    def mv(self, *args, **kwargs):
+        if self.worker:
+            self.fs.mv(*args, **kwargs)
+        else:
+            self.rfs.mv(*args, **kwargs).compute()
+
+    def ls(self, *args, **kwargs):
+        if self.worker:
+            return self.fs.ls(*args, **kwargs)
+        else:
+            return self.rfs.ls(*args, **kwargs).compute()
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        autocommit=True,
+        cache_options=None,
+        **kwargs,
+    ):
+        if self.worker:
+            return self.fs._open(
+                path,
+                mode=mode,
+                block_size=block_size,
+                autocommit=autocommit,
+                cache_options=cache_options,
+                **kwargs,
+            )
+        else:
+            return DaskFile(
+                fs=self,
+                path=path,
+                mode=mode,
+                block_size=block_size,
+                autocommit=autocommit,
+                cache_options=cache_options,
+                **kwargs,
+            )
+
+    def fetch_range(self, path, mode, start, end):
+        if self.worker:
+            with self._open(path, mode) as f:
+                f.seek(start)
+                return f.read(end - start)
+        else:
+            return self.rfs.fetch_range(path, mode, start, end).compute()
+
+
+class DaskFile(AbstractBufferedFile):
+    def __init__(self, mode="rb", **kwargs):
+        if mode != "rb":
+            raise ValueError('Remote dask files can only be opened in "rb" mode')
+        super().__init__(**kwargs)
+
+    def _upload_chunk(self, final=False):
+        pass
+
+    def _initiate_upload(self):
+        """Create remote file/upload"""
+        pass
+
+    def _fetch_range(self, start, end):
+        """Get the specified set of bytes from remote"""
+        return self.fs.fetch_range(self.path, self.mode, start, end)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/data.py

@@ -0,0 +1,58 @@
+import base64
+import io
+from typing import Optional
+from urllib.parse import unquote
+
+from fsspec import AbstractFileSystem
+
+
+class DataFileSystem(AbstractFileSystem):
+    """A handy decoder for data-URLs
+
+    Example
+    -------
+    >>> with fsspec.open("data:,Hello%2C%20World%21") as f:
+    ...     print(f.read())
+    b"Hello, World!"
+
+    See https://developer.mozilla.org/en-US/docs/Web/HTTP/Basics_of_HTTP/Data_URLs
+    """
+
+    protocol = "data"
+
+    def __init__(self, **kwargs):
+        """No parameters for this filesystem"""
+        super().__init__(**kwargs)
+
+    def cat_file(self, path, start=None, end=None, **kwargs):
+        pref, data = path.split(",", 1)
+        if pref.endswith("base64"):
+            return base64.b64decode(data)[start:end]
+        return unquote(data).encode()[start:end]
+
+    def info(self, path, **kwargs):
+        pref, name = path.split(",", 1)
+        data = self.cat_file(path)
+        mime = pref.split(":", 1)[1].split(";", 1)[0]
+        return {"name": name, "size": len(data), "type": "file", "mimetype": mime}
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        autocommit=True,
+        cache_options=None,
+        **kwargs,
+    ):
+        if "r" not in mode:
+            raise ValueError("Read only filesystem")
+        return io.BytesIO(self.cat_file(path))
+
+    @staticmethod
+    def encode(data: bytes, mime: Optional[str] = None):
+        """Format the given data into data-URL syntax
+
+        This version always base64 encodes, even when the data is ascii/url-safe.
+        """
+        return f"data:{mime or ''};base64,{base64.b64encode(data).decode()}"

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/dbfs.py

@@ -0,0 +1,496 @@
+from __future__ import annotations
+
+import base64
+import urllib
+
+import requests
+from requests.adapters import HTTPAdapter, Retry
+from typing_extensions import override
+
+from fsspec import AbstractFileSystem
+from fsspec.spec import AbstractBufferedFile
+
+
+class DatabricksException(Exception):
+    """
+    Helper class for exceptions raised in this module.
+    """
+
+    def __init__(self, error_code, message, details=None):
+        """Create a new DatabricksException"""
+        super().__init__(message)
+
+        self.error_code = error_code
+        self.message = message
+        self.details = details
+
+
+class DatabricksFileSystem(AbstractFileSystem):
+    """
+    Get access to the Databricks filesystem implementation over HTTP.
+    Can be used inside and outside of a databricks cluster.
+    """
+
+    def __init__(self, instance, token, **kwargs):
+        """
+        Create a new DatabricksFileSystem.
+
+        Parameters
+        ----------
+        instance: str
+            The instance URL of the databricks cluster.
+            For example for an Azure databricks cluster, this
+            has the form adb-<some-number>.<two digits>.azuredatabricks.net.
+        token: str
+            Your personal token. Find out more
+            here: https://docs.databricks.com/dev-tools/api/latest/authentication.html
+        """
+        self.instance = instance
+        self.token = token
+        self.session = requests.Session()
+        self.retries = Retry(
+            total=10,
+            backoff_factor=0.05,
+            status_forcelist=[408, 429, 500, 502, 503, 504],
+        )
+
+        self.session.mount("https://", HTTPAdapter(max_retries=self.retries))
+        self.session.headers.update({"Authorization": f"Bearer {self.token}"})
+
+        super().__init__(**kwargs)
+
+    @override
+    def _ls_from_cache(self, path) -> list[dict[str, str | int]] | None:
+        """Check cache for listing
+
+        Returns listing, if found (may be empty list for a directory that
+        exists but contains nothing), None if not in cache.
+        """
+        self.dircache.pop(path.rstrip("/"), None)
+
+        parent = self._parent(path)
+        if parent in self.dircache:
+            for entry in self.dircache[parent]:
+                if entry["name"] == path.rstrip("/"):
+                    if entry["type"] != "directory":
+                        return [entry]
+                    return []
+            raise FileNotFoundError(path)
+
+    def ls(self, path, detail=True, **kwargs):
+        """
+        List the contents of the given path.
+
+        Parameters
+        ----------
+        path: str
+            Absolute path
+        detail: bool
+            Return not only the list of filenames,
+            but also additional information on file sizes
+            and types.
+        """
+        try:
+            out = self._ls_from_cache(path)
+        except FileNotFoundError:
+            # This happens if the `path`'s parent was cached, but `path` is not
+            # there. This suggests that `path` is new since the parent was
+            # cached. Attempt to invalidate parent's cache before continuing.
+            self.dircache.pop(self._parent(path), None)
+            out = None
+
+        if not out:
+            try:
+                r = self._send_to_api(
+                    method="get", endpoint="list", json={"path": path}
+                )
+            except DatabricksException as e:
+                if e.error_code == "RESOURCE_DOES_NOT_EXIST":
+                    raise FileNotFoundError(e.message) from e
+
+                raise
+            files = r.get("files", [])
+            out = [
+                {
+                    "name": o["path"],
+                    "type": "directory" if o["is_dir"] else "file",
+                    "size": o["file_size"],
+                }
+                for o in files
+            ]
+            self.dircache[path] = out
+
+        if detail:
+            return out
+        return [o["name"] for o in out]
+
+    def makedirs(self, path, exist_ok=True):
+        """
+        Create a given absolute path and all of its parents.
+
+        Parameters
+        ----------
+        path: str
+            Absolute path to create
+        exist_ok: bool
+            If false, checks if the folder
+            exists before creating it (and raises an
+            Exception if this is the case)
+        """
+        if not exist_ok:
+            try:
+                # If the following succeeds, the path is already present
+                self._send_to_api(
+                    method="get", endpoint="get-status", json={"path": path}
+                )
+                raise FileExistsError(f"Path {path} already exists")
+            except DatabricksException as e:
+                if e.error_code == "RESOURCE_DOES_NOT_EXIST":
+                    pass
+
+        try:
+            self._send_to_api(method="post", endpoint="mkdirs", json={"path": path})
+        except DatabricksException as e:
+            if e.error_code == "RESOURCE_ALREADY_EXISTS":
+                raise FileExistsError(e.message) from e
+
+            raise
+        self.invalidate_cache(self._parent(path))
+
+    def mkdir(self, path, create_parents=True, **kwargs):
+        """
+        Create a given absolute path and all of its parents.
+
+        Parameters
+        ----------
+        path: str
+            Absolute path to create
+        create_parents: bool
+            Whether to create all parents or not.
+            "False" is not implemented so far.
+        """
+        if not create_parents:
+            raise NotImplementedError
+
+        self.mkdirs(path, **kwargs)
+
+    def rm(self, path, recursive=False, **kwargs):
+        """
+        Remove the file or folder at the given absolute path.
+
+        Parameters
+        ----------
+        path: str
+            Absolute path what to remove
+        recursive: bool
+            Recursively delete all files in a folder.
+        """
+        try:
+            self._send_to_api(
+                method="post",
+                endpoint="delete",
+                json={"path": path, "recursive": recursive},
+            )
+        except DatabricksException as e:
+            # This is not really an exception, it just means
+            # not everything was deleted so far
+            if e.error_code == "PARTIAL_DELETE":
+                self.rm(path=path, recursive=recursive)
+            elif e.error_code == "IO_ERROR":
+                # Using the same exception as the os module would use here
+                raise OSError(e.message) from e
+
+            raise
+        self.invalidate_cache(self._parent(path))
+
+    def mv(
+        self, source_path, destination_path, recursive=False, maxdepth=None, **kwargs
+    ):
+        """
+        Move a source to a destination path.
+
+        A note from the original [databricks API manual]
+        (https://docs.databricks.com/dev-tools/api/latest/dbfs.html#move).
+
+        When moving a large number of files the API call will time out after
+        approximately 60s, potentially resulting in partially moved data.
+        Therefore, for operations that move more than 10k files, we strongly
+        discourage using the DBFS REST API.
+
+        Parameters
+        ----------
+        source_path: str
+            From where to move (absolute path)
+        destination_path: str
+            To where to move (absolute path)
+        recursive: bool
+            Not implemented to far.
+        maxdepth:
+            Not implemented to far.
+        """
+        if recursive:
+            raise NotImplementedError
+        if maxdepth:
+            raise NotImplementedError
+
+        try:
+            self._send_to_api(
+                method="post",
+                endpoint="move",
+                json={"source_path": source_path, "destination_path": destination_path},
+            )
+        except DatabricksException as e:
+            if e.error_code == "RESOURCE_DOES_NOT_EXIST":
+                raise FileNotFoundError(e.message) from e
+            elif e.error_code == "RESOURCE_ALREADY_EXISTS":
+                raise FileExistsError(e.message) from e
+
+            raise
+        self.invalidate_cache(self._parent(source_path))
+        self.invalidate_cache(self._parent(destination_path))
+
+    def _open(self, path, mode="rb", block_size="default", **kwargs):
+        """
+        Overwrite the base class method to make sure to create a DBFile.
+        All arguments are copied from the base method.
+
+        Only the default blocksize is allowed.
+        """
+        return DatabricksFile(self, path, mode=mode, block_size=block_size, **kwargs)
+
+    def _send_to_api(self, method, endpoint, json):
+        """
+        Send the given json to the DBFS API
+        using a get or post request (specified by the argument `method`).
+
+        Parameters
+        ----------
+        method: str
+            Which http method to use for communication; "get" or "post".
+        endpoint: str
+            Where to send the request to (last part of the API URL)
+        json: dict
+            Dictionary of information to send
+        """
+        if method == "post":
+            session_call = self.session.post
+        elif method == "get":
+            session_call = self.session.get
+        else:
+            raise ValueError(f"Do not understand method {method}")
+
+        url = urllib.parse.urljoin(f"https://{self.instance}/api/2.0/dbfs/", endpoint)
+
+        r = session_call(url, json=json)
+
+        # The DBFS API will return a json, also in case of an exception.
+        # We want to preserve this information as good as possible.
+        try:
+            r.raise_for_status()
+        except requests.HTTPError as e:
+            # try to extract json error message
+            # if that fails, fall back to the original exception
+            try:
+                exception_json = e.response.json()
+            except Exception:
+                raise e from None
+
+            raise DatabricksException(**exception_json) from e
+
+        return r.json()
+
+    def _create_handle(self, path, overwrite=True):
+        """
+        Internal function to create a handle, which can be used to
+        write blocks of a file to DBFS.
+        A handle has a unique identifier which needs to be passed
+        whenever written during this transaction.
+        The handle is active for 10 minutes - after that a new
+        write transaction needs to be created.
+        Make sure to close the handle after you are finished.
+
+        Parameters
+        ----------
+        path: str
+            Absolute path for this file.
+        overwrite: bool
+            If a file already exist at this location, either overwrite
+            it or raise an exception.
+        """
+        try:
+            r = self._send_to_api(
+                method="post",
+                endpoint="create",
+                json={"path": path, "overwrite": overwrite},
+            )
+            return r["handle"]
+        except DatabricksException as e:
+            if e.error_code == "RESOURCE_ALREADY_EXISTS":
+                raise FileExistsError(e.message) from e
+
+            raise
+
+    def _close_handle(self, handle):
+        """
+        Close a handle, which was opened by :func:`_create_handle`.
+
+        Parameters
+        ----------
+        handle: str
+            Which handle to close.
+        """
+        try:
+            self._send_to_api(method="post", endpoint="close", json={"handle": handle})
+        except DatabricksException as e:
+            if e.error_code == "RESOURCE_DOES_NOT_EXIST":
+                raise FileNotFoundError(e.message) from e
+
+            raise
+
+    def _add_data(self, handle, data):
+        """
+        Upload data to an already opened file handle
+        (opened by :func:`_create_handle`).
+        The maximal allowed data size is 1MB after
+        conversion to base64.
+        Remember to close the handle when you are finished.
+
+        Parameters
+        ----------
+        handle: str
+            Which handle to upload data to.
+        data: bytes
+            Block of data to add to the handle.
+        """
+        data = base64.b64encode(data).decode()
+        try:
+            self._send_to_api(
+                method="post",
+                endpoint="add-block",
+                json={"handle": handle, "data": data},
+            )
+        except DatabricksException as e:
+            if e.error_code == "RESOURCE_DOES_NOT_EXIST":
+                raise FileNotFoundError(e.message) from e
+            elif e.error_code == "MAX_BLOCK_SIZE_EXCEEDED":
+                raise ValueError(e.message) from e
+
+            raise
+
+    def _get_data(self, path, start, end):
+        """
+        Download data in bytes from a given absolute path in a block
+        from [start, start+length].
+        The maximum number of allowed bytes to read is 1MB.
+
+        Parameters
+        ----------
+        path: str
+            Absolute path to download data from
+        start: int
+            Start position of the block
+        end: int
+            End position of the block
+        """
+        try:
+            r = self._send_to_api(
+                method="get",
+                endpoint="read",
+                json={"path": path, "offset": start, "length": end - start},
+            )
+            return base64.b64decode(r["data"])
+        except DatabricksException as e:
+            if e.error_code == "RESOURCE_DOES_NOT_EXIST":
+                raise FileNotFoundError(e.message) from e
+            elif e.error_code in ["INVALID_PARAMETER_VALUE", "MAX_READ_SIZE_EXCEEDED"]:
+                raise ValueError(e.message) from e
+
+            raise
+
+    def invalidate_cache(self, path=None):
+        if path is None:
+            self.dircache.clear()
+        else:
+            self.dircache.pop(path, None)
+        super().invalidate_cache(path)
+
+
+class DatabricksFile(AbstractBufferedFile):
+    """
+    Helper class for files referenced in the DatabricksFileSystem.
+    """
+
+    DEFAULT_BLOCK_SIZE = 1 * 2**20  # only allowed block size
+
+    def __init__(
+        self,
+        fs,
+        path,
+        mode="rb",
+        block_size="default",
+        autocommit=True,
+        cache_type="readahead",
+        cache_options=None,
+        **kwargs,
+    ):
+        """
+        Create a new instance of the DatabricksFile.
+
+        The blocksize needs to be the default one.
+        """
+        if block_size is None or block_size == "default":
+            block_size = self.DEFAULT_BLOCK_SIZE
+
+        assert block_size == self.DEFAULT_BLOCK_SIZE, (
+            f"Only the default block size is allowed, not {block_size}"
+        )
+
+        super().__init__(
+            fs,
+            path,
+            mode=mode,
+            block_size=block_size,
+            autocommit=autocommit,
+            cache_type=cache_type,
+            cache_options=cache_options or {},
+            **kwargs,
+        )
+
+    def _initiate_upload(self):
+        """Internal function to start a file upload"""
+        self.handle = self.fs._create_handle(self.path)
+
+    def _upload_chunk(self, final=False):
+        """Internal function to add a chunk of data to a started upload"""
+        self.buffer.seek(0)
+        data = self.buffer.getvalue()
+
+        data_chunks = [
+            data[start:end] for start, end in self._to_sized_blocks(len(data))
+        ]
+
+        for data_chunk in data_chunks:
+            self.fs._add_data(handle=self.handle, data=data_chunk)
+
+        if final:
+            self.fs._close_handle(handle=self.handle)
+            return True
+
+    def _fetch_range(self, start, end):
+        """Internal function to download a block of data"""
+        return_buffer = b""
+        length = end - start
+        for chunk_start, chunk_end in self._to_sized_blocks(length, start):
+            return_buffer += self.fs._get_data(
+                path=self.path, start=chunk_start, end=chunk_end
+            )
+
+        return return_buffer
+
+    def _to_sized_blocks(self, length, start=0):
+        """Helper function to split a range from 0 to total_length into blocksizes"""
+        end = start + length
+        for data_chunk in range(start, end, self.blocksize):
+            data_start = data_chunk
+            data_end = min(end, data_chunk + self.blocksize)
+            yield data_start, data_end

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/dirfs.py

@@ -0,0 +1,388 @@
+from .. import filesystem
+from ..asyn import AsyncFileSystem
+
+
+class DirFileSystem(AsyncFileSystem):
+    """Directory prefix filesystem
+
+    The DirFileSystem is a filesystem-wrapper. It assumes every path it is dealing with
+    is relative to the `path`. After performing the necessary paths operation it
+    delegates everything to the wrapped filesystem.
+    """
+
+    protocol = "dir"
+
+    def __init__(
+        self,
+        path=None,
+        fs=None,
+        fo=None,
+        target_protocol=None,
+        target_options=None,
+        **storage_options,
+    ):
+        """
+        Parameters
+        ----------
+        path: str
+            Path to the directory.
+        fs: AbstractFileSystem
+            An instantiated filesystem to wrap.
+        target_protocol, target_options:
+            if fs is none, construct it from these
+        fo: str
+            Alternate for path; do not provide both
+        """
+        super().__init__(**storage_options)
+        if fs is None:
+            fs = filesystem(protocol=target_protocol, **(target_options or {}))
+        path = path or fo
+
+        if self.asynchronous and not fs.async_impl:
+            raise ValueError("can't use asynchronous with non-async fs")
+
+        if fs.async_impl and self.asynchronous != fs.asynchronous:
+            raise ValueError("both dirfs and fs should be in the same sync/async mode")
+
+        self.path = fs._strip_protocol(path)
+        self.fs = fs
+
+    def _join(self, path):
+        if isinstance(path, str):
+            if not self.path:
+                return path
+            if not path:
+                return self.path
+            return self.fs.sep.join((self.path, self._strip_protocol(path)))
+        if isinstance(path, dict):
+            return {self._join(_path): value for _path, value in path.items()}
+        return [self._join(_path) for _path in path]
+
+    def _relpath(self, path):
+        if isinstance(path, str):
+            if not self.path:
+                return path
+            # We need to account for S3FileSystem returning paths that do not
+            # start with a '/'
+            if path == self.path or (
+                self.path.startswith(self.fs.sep) and path == self.path[1:]
+            ):
+                return ""
+            prefix = self.path + self.fs.sep
+            if self.path.startswith(self.fs.sep) and not path.startswith(self.fs.sep):
+                prefix = prefix[1:]
+            assert path.startswith(prefix)
+            return path[len(prefix) :]
+        return [self._relpath(_path) for _path in path]
+
+    # Wrappers below
+
+    @property
+    def sep(self):
+        return self.fs.sep
+
+    async def set_session(self, *args, **kwargs):
+        return await self.fs.set_session(*args, **kwargs)
+
+    async def _rm_file(self, path, **kwargs):
+        return await self.fs._rm_file(self._join(path), **kwargs)
+
+    def rm_file(self, path, **kwargs):
+        return self.fs.rm_file(self._join(path), **kwargs)
+
+    async def _rm(self, path, *args, **kwargs):
+        return await self.fs._rm(self._join(path), *args, **kwargs)
+
+    def rm(self, path, *args, **kwargs):
+        return self.fs.rm(self._join(path), *args, **kwargs)
+
+    async def _cp_file(self, path1, path2, **kwargs):
+        return await self.fs._cp_file(self._join(path1), self._join(path2), **kwargs)
+
+    def cp_file(self, path1, path2, **kwargs):
+        return self.fs.cp_file(self._join(path1), self._join(path2), **kwargs)
+
+    async def _copy(
+        self,
+        path1,
+        path2,
+        *args,
+        **kwargs,
+    ):
+        return await self.fs._copy(
+            self._join(path1),
+            self._join(path2),
+            *args,
+            **kwargs,
+        )
+
+    def copy(self, path1, path2, *args, **kwargs):
+        return self.fs.copy(
+            self._join(path1),
+            self._join(path2),
+            *args,
+            **kwargs,
+        )
+
+    async def _pipe(self, path, *args, **kwargs):
+        return await self.fs._pipe(self._join(path), *args, **kwargs)
+
+    def pipe(self, path, *args, **kwargs):
+        return self.fs.pipe(self._join(path), *args, **kwargs)
+
+    async def _pipe_file(self, path, *args, **kwargs):
+        return await self.fs._pipe_file(self._join(path), *args, **kwargs)
+
+    def pipe_file(self, path, *args, **kwargs):
+        return self.fs.pipe_file(self._join(path), *args, **kwargs)
+
+    async def _cat_file(self, path, *args, **kwargs):
+        return await self.fs._cat_file(self._join(path), *args, **kwargs)
+
+    def cat_file(self, path, *args, **kwargs):
+        return self.fs.cat_file(self._join(path), *args, **kwargs)
+
+    async def _cat(self, path, *args, **kwargs):
+        ret = await self.fs._cat(
+            self._join(path),
+            *args,
+            **kwargs,
+        )
+
+        if isinstance(ret, dict):
+            return {self._relpath(key): value for key, value in ret.items()}
+
+        return ret
+
+    def cat(self, path, *args, **kwargs):
+        ret = self.fs.cat(
+            self._join(path),
+            *args,
+            **kwargs,
+        )
+
+        if isinstance(ret, dict):
+            return {self._relpath(key): value for key, value in ret.items()}
+
+        return ret
+
+    async def _put_file(self, lpath, rpath, **kwargs):
+        return await self.fs._put_file(lpath, self._join(rpath), **kwargs)
+
+    def put_file(self, lpath, rpath, **kwargs):
+        return self.fs.put_file(lpath, self._join(rpath), **kwargs)
+
+    async def _put(
+        self,
+        lpath,
+        rpath,
+        *args,
+        **kwargs,
+    ):
+        return await self.fs._put(
+            lpath,
+            self._join(rpath),
+            *args,
+            **kwargs,
+        )
+
+    def put(self, lpath, rpath, *args, **kwargs):
+        return self.fs.put(
+            lpath,
+            self._join(rpath),
+            *args,
+            **kwargs,
+        )
+
+    async def _get_file(self, rpath, lpath, **kwargs):
+        return await self.fs._get_file(self._join(rpath), lpath, **kwargs)
+
+    def get_file(self, rpath, lpath, **kwargs):
+        return self.fs.get_file(self._join(rpath), lpath, **kwargs)
+
+    async def _get(self, rpath, *args, **kwargs):
+        return await self.fs._get(self._join(rpath), *args, **kwargs)
+
+    def get(self, rpath, *args, **kwargs):
+        return self.fs.get(self._join(rpath), *args, **kwargs)
+
+    async def _isfile(self, path):
+        return await self.fs._isfile(self._join(path))
+
+    def isfile(self, path):
+        return self.fs.isfile(self._join(path))
+
+    async def _isdir(self, path):
+        return await self.fs._isdir(self._join(path))
+
+    def isdir(self, path):
+        return self.fs.isdir(self._join(path))
+
+    async def _size(self, path):
+        return await self.fs._size(self._join(path))
+
+    def size(self, path):
+        return self.fs.size(self._join(path))
+
+    async def _exists(self, path):
+        return await self.fs._exists(self._join(path))
+
+    def exists(self, path):
+        return self.fs.exists(self._join(path))
+
+    async def _info(self, path, **kwargs):
+        info = await self.fs._info(self._join(path), **kwargs)
+        info = info.copy()
+        info["name"] = self._relpath(info["name"])
+        return info
+
+    def info(self, path, **kwargs):
+        info = self.fs.info(self._join(path), **kwargs)
+        info = info.copy()
+        info["name"] = self._relpath(info["name"])
+        return info
+
+    async def _ls(self, path, detail=True, **kwargs):
+        ret = (await self.fs._ls(self._join(path), detail=detail, **kwargs)).copy()
+        if detail:
+            out = []
+            for entry in ret:
+                entry = entry.copy()
+                entry["name"] = self._relpath(entry["name"])
+                out.append(entry)
+            return out
+
+        return self._relpath(ret)
+
+    def ls(self, path, detail=True, **kwargs):
+        ret = self.fs.ls(self._join(path), detail=detail, **kwargs).copy()
+        if detail:
+            out = []
+            for entry in ret:
+                entry = entry.copy()
+                entry["name"] = self._relpath(entry["name"])
+                out.append(entry)
+            return out
+
+        return self._relpath(ret)
+
+    async def _walk(self, path, *args, **kwargs):
+        async for root, dirs, files in self.fs._walk(self._join(path), *args, **kwargs):
+            yield self._relpath(root), dirs, files
+
+    def walk(self, path, *args, **kwargs):
+        for root, dirs, files in self.fs.walk(self._join(path), *args, **kwargs):
+            yield self._relpath(root), dirs, files
+
+    async def _glob(self, path, **kwargs):
+        detail = kwargs.get("detail", False)
+        ret = await self.fs._glob(self._join(path), **kwargs)
+        if detail:
+            return {self._relpath(path): info for path, info in ret.items()}
+        return self._relpath(ret)
+
+    def glob(self, path, **kwargs):
+        detail = kwargs.get("detail", False)
+        ret = self.fs.glob(self._join(path), **kwargs)
+        if detail:
+            return {self._relpath(path): info for path, info in ret.items()}
+        return self._relpath(ret)
+
+    async def _du(self, path, *args, **kwargs):
+        total = kwargs.get("total", True)
+        ret = await self.fs._du(self._join(path), *args, **kwargs)
+        if total:
+            return ret
+
+        return {self._relpath(path): size for path, size in ret.items()}
+
+    def du(self, path, *args, **kwargs):
+        total = kwargs.get("total", True)
+        ret = self.fs.du(self._join(path), *args, **kwargs)
+        if total:
+            return ret
+
+        return {self._relpath(path): size for path, size in ret.items()}
+
+    async def _find(self, path, *args, **kwargs):
+        detail = kwargs.get("detail", False)
+        ret = await self.fs._find(self._join(path), *args, **kwargs)
+        if detail:
+            return {self._relpath(path): info for path, info in ret.items()}
+        return self._relpath(ret)
+
+    def find(self, path, *args, **kwargs):
+        detail = kwargs.get("detail", False)
+        ret = self.fs.find(self._join(path), *args, **kwargs)
+        if detail:
+            return {self._relpath(path): info for path, info in ret.items()}
+        return self._relpath(ret)
+
+    async def _expand_path(self, path, *args, **kwargs):
+        return self._relpath(
+            await self.fs._expand_path(self._join(path), *args, **kwargs)
+        )
+
+    def expand_path(self, path, *args, **kwargs):
+        return self._relpath(self.fs.expand_path(self._join(path), *args, **kwargs))
+
+    async def _mkdir(self, path, *args, **kwargs):
+        return await self.fs._mkdir(self._join(path), *args, **kwargs)
+
+    def mkdir(self, path, *args, **kwargs):
+        return self.fs.mkdir(self._join(path), *args, **kwargs)
+
+    async def _makedirs(self, path, *args, **kwargs):
+        return await self.fs._makedirs(self._join(path), *args, **kwargs)
+
+    def makedirs(self, path, *args, **kwargs):
+        return self.fs.makedirs(self._join(path), *args, **kwargs)
+
+    def rmdir(self, path):
+        return self.fs.rmdir(self._join(path))
+
+    def mv(self, path1, path2, **kwargs):
+        return self.fs.mv(
+            self._join(path1),
+            self._join(path2),
+            **kwargs,
+        )
+
+    def touch(self, path, **kwargs):
+        return self.fs.touch(self._join(path), **kwargs)
+
+    def created(self, path):
+        return self.fs.created(self._join(path))
+
+    def modified(self, path):
+        return self.fs.modified(self._join(path))
+
+    def sign(self, path, *args, **kwargs):
+        return self.fs.sign(self._join(path), *args, **kwargs)
+
+    def __repr__(self):
+        return f"{self.__class__.__qualname__}(path='{self.path}', fs={self.fs})"
+
+    def open(
+        self,
+        path,
+        *args,
+        **kwargs,
+    ):
+        return self.fs.open(
+            self._join(path),
+            *args,
+            **kwargs,
+        )
+
+    async def open_async(
+        self,
+        path,
+        *args,
+        **kwargs,
+    ):
+        return await self.fs.open_async(
+            self._join(path),
+            *args,
+            **kwargs,
+        )

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/ftp.py

@@ -0,0 +1,387 @@
+import os
+import uuid
+from ftplib import FTP, FTP_TLS, Error, error_perm
+from typing import Any
+
+from ..spec import AbstractBufferedFile, AbstractFileSystem
+from ..utils import infer_storage_options, isfilelike
+
+
+class FTPFileSystem(AbstractFileSystem):
+    """A filesystem over classic FTP"""
+
+    root_marker = "/"
+    cachable = False
+    protocol = "ftp"
+
+    def __init__(
+        self,
+        host,
+        port=21,
+        username=None,
+        password=None,
+        acct=None,
+        block_size=None,
+        tempdir=None,
+        timeout=30,
+        encoding="utf-8",
+        tls=False,
+        **kwargs,
+    ):
+        """
+        You can use _get_kwargs_from_urls to get some kwargs from
+        a reasonable FTP url.
+
+        Authentication will be anonymous if username/password are not
+        given.
+
+        Parameters
+        ----------
+        host: str
+            The remote server name/ip to connect to
+        port: int
+            Port to connect with
+        username: str or None
+            If authenticating, the user's identifier
+        password: str of None
+            User's password on the server, if using
+        acct: str or None
+            Some servers also need an "account" string for auth
+        block_size: int or None
+            If given, the read-ahead or write buffer size.
+        tempdir: str
+            Directory on remote to put temporary files when in a transaction
+        timeout: int
+            Timeout of the ftp connection in seconds
+        encoding: str
+            Encoding to use for directories and filenames in FTP connection
+        tls: bool
+            Use FTP-TLS, by default False
+        """
+        super().__init__(**kwargs)
+        self.host = host
+        self.port = port
+        self.tempdir = tempdir or "/tmp"
+        self.cred = username or "", password or "", acct or ""
+        self.timeout = timeout
+        self.encoding = encoding
+        if block_size is not None:
+            self.blocksize = block_size
+        else:
+            self.blocksize = 2**16
+        self.tls = tls
+        self._connect()
+        if self.tls:
+            self.ftp.prot_p()
+
+    def _connect(self):
+        if self.tls:
+            ftp_cls = FTP_TLS
+        else:
+            ftp_cls = FTP
+        self.ftp = ftp_cls(timeout=self.timeout, encoding=self.encoding)
+        self.ftp.connect(self.host, self.port)
+        self.ftp.login(*self.cred)
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        return "/" + infer_storage_options(path)["path"].lstrip("/").rstrip("/")
+
+    @staticmethod
+    def _get_kwargs_from_urls(urlpath):
+        out = infer_storage_options(urlpath)
+        out.pop("path", None)
+        out.pop("protocol", None)
+        return out
+
+    def ls(self, path, detail=True, **kwargs):
+        path = self._strip_protocol(path)
+        out = []
+        if path not in self.dircache:
+            try:
+                try:
+                    out = [
+                        (fn, details)
+                        for (fn, details) in self.ftp.mlsd(path)
+                        if fn not in [".", ".."]
+                        and details["type"] not in ["pdir", "cdir"]
+                    ]
+                except error_perm:
+                    out = _mlsd2(self.ftp, path)  # Not platform independent
+                for fn, details in out:
+                    details["name"] = "/".join(
+                        ["" if path == "/" else path, fn.lstrip("/")]
+                    )
+                    if details["type"] == "file":
+                        details["size"] = int(details["size"])
+                    else:
+                        details["size"] = 0
+                    if details["type"] == "dir":
+                        details["type"] = "directory"
+                self.dircache[path] = out
+            except Error:
+                try:
+                    info = self.info(path)
+                    if info["type"] == "file":
+                        out = [(path, info)]
+                except (Error, IndexError) as exc:
+                    raise FileNotFoundError(path) from exc
+        files = self.dircache.get(path, out)
+        if not detail:
+            return sorted([fn for fn, details in files])
+        return [details for fn, details in files]
+
+    def info(self, path, **kwargs):
+        # implement with direct method
+        path = self._strip_protocol(path)
+        if path == "/":
+            # special case, since this dir has no real entry
+            return {"name": "/", "size": 0, "type": "directory"}
+        files = self.ls(self._parent(path).lstrip("/"), True)
+        try:
+            out = next(f for f in files if f["name"] == path)
+        except StopIteration as exc:
+            raise FileNotFoundError(path) from exc
+        return out
+
+    def get_file(self, rpath, lpath, **kwargs):
+        if self.isdir(rpath):
+            if not os.path.exists(lpath):
+                os.mkdir(lpath)
+            return
+        if isfilelike(lpath):
+            outfile = lpath
+        else:
+            outfile = open(lpath, "wb")
+
+        def cb(x):
+            outfile.write(x)
+
+        self.ftp.retrbinary(
+            f"RETR {rpath}",
+            blocksize=self.blocksize,
+            callback=cb,
+        )
+        if not isfilelike(lpath):
+            outfile.close()
+
+    def cat_file(self, path, start=None, end=None, **kwargs):
+        if end is not None:
+            return super().cat_file(path, start, end, **kwargs)
+        out = []
+
+        def cb(x):
+            out.append(x)
+
+        try:
+            self.ftp.retrbinary(
+                f"RETR {path}",
+                blocksize=self.blocksize,
+                rest=start,
+                callback=cb,
+            )
+        except (Error, error_perm) as orig_exc:
+            raise FileNotFoundError(path) from orig_exc
+        return b"".join(out)
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        cache_options=None,
+        autocommit=True,
+        **kwargs,
+    ):
+        path = self._strip_protocol(path)
+        block_size = block_size or self.blocksize
+        return FTPFile(
+            self,
+            path,
+            mode=mode,
+            block_size=block_size,
+            tempdir=self.tempdir,
+            autocommit=autocommit,
+            cache_options=cache_options,
+        )
+
+    def _rm(self, path):
+        path = self._strip_protocol(path)
+        self.ftp.delete(path)
+        self.invalidate_cache(self._parent(path))
+
+    def rm(self, path, recursive=False, maxdepth=None):
+        paths = self.expand_path(path, recursive=recursive, maxdepth=maxdepth)
+        for p in reversed(paths):
+            if self.isfile(p):
+                self.rm_file(p)
+            else:
+                self.rmdir(p)
+
+    def mkdir(self, path: str, create_parents: bool = True, **kwargs: Any) -> None:
+        path = self._strip_protocol(path)
+        parent = self._parent(path)
+        if parent != self.root_marker and not self.exists(parent) and create_parents:
+            self.mkdir(parent, create_parents=create_parents)
+
+        self.ftp.mkd(path)
+        self.invalidate_cache(self._parent(path))
+
+    def makedirs(self, path: str, exist_ok: bool = False) -> None:
+        path = self._strip_protocol(path)
+        if self.exists(path):
+            # NB: "/" does not "exist" as it has no directory entry
+            if not exist_ok:
+                raise FileExistsError(f"{path} exists without `exist_ok`")
+            # exists_ok=True -> no-op
+        else:
+            self.mkdir(path, create_parents=True)
+
+    def rmdir(self, path):
+        path = self._strip_protocol(path)
+        self.ftp.rmd(path)
+        self.invalidate_cache(self._parent(path))
+
+    def mv(self, path1, path2, **kwargs):
+        path1 = self._strip_protocol(path1)
+        path2 = self._strip_protocol(path2)
+        self.ftp.rename(path1, path2)
+        self.invalidate_cache(self._parent(path1))
+        self.invalidate_cache(self._parent(path2))
+
+    def __del__(self):
+        self.ftp.close()
+
+    def invalidate_cache(self, path=None):
+        if path is None:
+            self.dircache.clear()
+        else:
+            self.dircache.pop(path, None)
+        super().invalidate_cache(path)
+
+
+class TransferDone(Exception):
+    """Internal exception to break out of transfer"""
+
+    pass
+
+
+class FTPFile(AbstractBufferedFile):
+    """Interact with a remote FTP file with read/write buffering"""
+
+    def __init__(
+        self,
+        fs,
+        path,
+        mode="rb",
+        block_size="default",
+        autocommit=True,
+        cache_type="readahead",
+        cache_options=None,
+        **kwargs,
+    ):
+        super().__init__(
+            fs,
+            path,
+            mode=mode,
+            block_size=block_size,
+            autocommit=autocommit,
+            cache_type=cache_type,
+            cache_options=cache_options,
+            **kwargs,
+        )
+        if not autocommit:
+            self.target = self.path
+            self.path = "/".join([kwargs["tempdir"], str(uuid.uuid4())])
+
+    def commit(self):
+        self.fs.mv(self.path, self.target)
+
+    def discard(self):
+        self.fs.rm(self.path)
+
+    def _fetch_range(self, start, end):
+        """Get bytes between given byte limits
+
+        Implemented by raising an exception in the fetch callback when the
+        number of bytes received reaches the requested amount.
+
+        Will fail if the server does not respect the REST command on
+        retrieve requests.
+        """
+        out = []
+        total = [0]
+
+        def callback(x):
+            total[0] += len(x)
+            if total[0] > end - start:
+                out.append(x[: (end - start) - total[0]])
+                if end < self.size:
+                    raise TransferDone
+            else:
+                out.append(x)
+
+            if total[0] == end - start and end < self.size:
+                raise TransferDone
+
+        try:
+            self.fs.ftp.retrbinary(
+                f"RETR {self.path}",
+                blocksize=self.blocksize,
+                rest=start,
+                callback=callback,
+            )
+        except TransferDone:
+            try:
+                # stop transfer, we got enough bytes for this block
+                self.fs.ftp.abort()
+                self.fs.ftp.getmultiline()
+            except Error:
+                self.fs._connect()
+
+        return b"".join(out)
+
+    def _upload_chunk(self, final=False):
+        self.buffer.seek(0)
+        self.fs.ftp.storbinary(
+            f"STOR {self.path}", self.buffer, blocksize=self.blocksize, rest=self.offset
+        )
+        return True
+
+
+def _mlsd2(ftp, path="."):
+    """
+    Fall back to using `dir` instead of `mlsd` if not supported.
+
+    This parses a Linux style `ls -l` response to `dir`, but the response may
+    be platform dependent.
+
+    Parameters
+    ----------
+    ftp: ftplib.FTP
+    path: str
+        Expects to be given path, but defaults to ".".
+    """
+    lines = []
+    minfo = []
+    ftp.dir(path, lines.append)
+    for line in lines:
+        split_line = line.split()
+        if len(split_line) < 9:
+            continue
+        this = (
+            split_line[-1],
+            {
+                "modify": " ".join(split_line[5:8]),
+                "unix.owner": split_line[2],
+                "unix.group": split_line[3],
+                "unix.mode": split_line[0],
+                "size": split_line[4],
+            },
+        )
+        if this[1]["unix.mode"][0] == "d":
+            this[1]["type"] = "dir"
+        else:
+            this[1]["type"] = "file"
+        minfo.append(this)
+    return minfo

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/gist.py

@@ -0,0 +1,241 @@
+import requests
+
+from ..spec import AbstractFileSystem
+from ..utils import infer_storage_options
+from .memory import MemoryFile
+
+
+class GistFileSystem(AbstractFileSystem):
+    """
+    Interface to files in a single GitHub Gist.
+
+    Provides read-only access to a gist's files. Gists do not contain
+    subdirectories, so file listing is straightforward.
+
+    Parameters
+    ----------
+    gist_id: str
+        The ID of the gist you want to access (the long hex value from the URL).
+    filenames: list[str] (optional)
+        If provided, only make a file system representing these files, and do not fetch
+        the list of all files for this gist.
+    sha: str (optional)
+        If provided, fetch a particular revision of the gist. If omitted,
+        the latest revision is used.
+    username: str (optional)
+        GitHub username for authentication.
+    token: str (optional)
+        GitHub personal access token (required if username is given), or.
+    timeout: (float, float) or float, optional
+        Connect and read timeouts for requests (default 60s each).
+    kwargs: dict
+        Stored on `self.request_kw` and passed to `requests.get` when fetching Gist
+        metadata or reading ("opening") a file.
+    """
+
+    protocol = "gist"
+    gist_url = "https://api.github.com/gists/{gist_id}"
+    gist_rev_url = "https://api.github.com/gists/{gist_id}/{sha}"
+
+    def __init__(
+        self,
+        gist_id,
+        filenames=None,
+        sha=None,
+        username=None,
+        token=None,
+        timeout=None,
+        **kwargs,
+    ):
+        super().__init__()
+        self.gist_id = gist_id
+        self.filenames = filenames
+        self.sha = sha  # revision of the gist (optional)
+        if username is not None and token is None:
+            raise ValueError("User auth requires a token")
+        self.username = username
+        self.token = token
+        self.request_kw = kwargs
+        # Default timeouts to 60s connect/read if none provided
+        self.timeout = timeout if timeout is not None else (60, 60)
+
+        # We use a single-level "directory" cache, because a gist is essentially flat
+        self.dircache[""] = self._fetch_file_list()
+
+    @property
+    def kw(self):
+        """Auth parameters passed to 'requests' if we have username/token."""
+        kw = {
+            "headers": {
+                "Accept": "application/vnd.github+json",
+                "X-GitHub-Api-Version": "2022-11-28",
+            }
+        }
+        kw.update(self.request_kw)
+        if self.username and self.token:
+            kw["auth"] = (self.username, self.token)
+        elif self.token:
+            kw["headers"]["Authorization"] = f"Bearer {self.token}"
+        return kw
+
+    def _fetch_gist_metadata(self):
+        """
+        Fetch the JSON metadata for this gist (possibly for a specific revision).
+        """
+        if self.sha:
+            url = self.gist_rev_url.format(gist_id=self.gist_id, sha=self.sha)
+        else:
+            url = self.gist_url.format(gist_id=self.gist_id)
+
+        r = requests.get(url, timeout=self.timeout, **self.kw)
+        if r.status_code == 404:
+            raise FileNotFoundError(
+                f"Gist not found: {self.gist_id}@{self.sha or 'latest'}"
+            )
+        r.raise_for_status()
+        return r.json()
+
+    def _fetch_file_list(self):
+        """
+        Returns a list of dicts describing each file in the gist. These get stored
+        in self.dircache[""].
+        """
+        meta = self._fetch_gist_metadata()
+        if self.filenames:
+            available_files = meta.get("files", {})
+            files = {}
+            for fn in self.filenames:
+                if fn not in available_files:
+                    raise FileNotFoundError(fn)
+                files[fn] = available_files[fn]
+        else:
+            files = meta.get("files", {})
+
+        out = []
+        for fname, finfo in files.items():
+            if finfo is None:
+                # Occasionally GitHub returns a file entry with null if it was deleted
+                continue
+            # Build a directory entry
+            out.append(
+                {
+                    "name": fname,  # file's name
+                    "type": "file",  # gists have no subdirectories
+                    "size": finfo.get("size", 0),  # file size in bytes
+                    "raw_url": finfo.get("raw_url"),
+                }
+            )
+        return out
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        """
+        Remove 'gist://' from the path, if present.
+        """
+        # The default infer_storage_options can handle gist://username:token@id/file
+        # or gist://id/file, but let's ensure we handle a normal usage too.
+        # We'll just strip the protocol prefix if it exists.
+        path = infer_storage_options(path).get("path", path)
+        return path.lstrip("/")
+
+    @staticmethod
+    def _get_kwargs_from_urls(path):
+        """
+        Parse 'gist://' style URLs into GistFileSystem constructor kwargs.
+        For example:
+          gist://:TOKEN@<gist_id>/file.txt
+          gist://username:TOKEN@<gist_id>/file.txt
+        """
+        so = infer_storage_options(path)
+        out = {}
+        if "username" in so and so["username"]:
+            out["username"] = so["username"]
+        if "password" in so and so["password"]:
+            out["token"] = so["password"]
+        if "host" in so and so["host"]:
+            # We interpret 'host' as the gist ID
+            out["gist_id"] = so["host"]
+
+        # Extract SHA and filename from path
+        if "path" in so and so["path"]:
+            path_parts = so["path"].rsplit("/", 2)[-2:]
+            if len(path_parts) == 2:
+                if path_parts[0]:  # SHA present
+                    out["sha"] = path_parts[0]
+                if path_parts[1]:  # filename also present
+                    out["filenames"] = [path_parts[1]]
+
+        return out
+
+    def ls(self, path="", detail=False, **kwargs):
+        """
+        List files in the gist. Gists are single-level, so any 'path' is basically
+        the filename, or empty for all files.
+
+        Parameters
+        ----------
+        path : str, optional
+            The filename to list. If empty, returns all files in the gist.
+        detail : bool, default False
+            If True, return a list of dicts; if False, return a list of filenames.
+        """
+        path = self._strip_protocol(path or "")
+        # If path is empty, return all
+        if path == "":
+            results = self.dircache[""]
+        else:
+            # We want just the single file with this name
+            all_files = self.dircache[""]
+            results = [f for f in all_files if f["name"] == path]
+            if not results:
+                raise FileNotFoundError(path)
+        if detail:
+            return results
+        else:
+            return sorted(f["name"] for f in results)
+
+    def _open(self, path, mode="rb", block_size=None, **kwargs):
+        """
+        Read a single file from the gist.
+        """
+        if mode != "rb":
+            raise NotImplementedError("GitHub Gist FS is read-only (no write).")
+
+        path = self._strip_protocol(path)
+        # Find the file entry in our dircache
+        matches = [f for f in self.dircache[""] if f["name"] == path]
+        if not matches:
+            raise FileNotFoundError(path)
+        finfo = matches[0]
+
+        raw_url = finfo.get("raw_url")
+        if not raw_url:
+            raise FileNotFoundError(f"No raw_url for file: {path}")
+
+        r = requests.get(raw_url, timeout=self.timeout, **self.kw)
+        if r.status_code == 404:
+            raise FileNotFoundError(path)
+        r.raise_for_status()
+        return MemoryFile(path, None, r.content)
+
+    def cat(self, path, recursive=False, on_error="raise", **kwargs):
+        """
+        Return {path: contents} for the given file or files. If 'recursive' is True,
+        and path is empty, returns all files in the gist.
+        """
+        paths = self.expand_path(path, recursive=recursive)
+        out = {}
+        for p in paths:
+            try:
+                with self.open(p, "rb") as f:
+                    out[p] = f.read()
+            except FileNotFoundError as e:
+                if on_error == "raise":
+                    raise e
+                elif on_error == "omit":
+                    pass  # skip
+                else:
+                    out[p] = e
+        if len(paths) == 1 and paths[0] == path:
+            return out[path]
+        return out

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/git.py

@@ -0,0 +1,114 @@
+import os
+
+import pygit2
+
+from fsspec.spec import AbstractFileSystem
+
+from .memory import MemoryFile
+
+
+class GitFileSystem(AbstractFileSystem):
+    """Browse the files of a local git repo at any hash/tag/branch
+
+    (experimental backend)
+    """
+
+    root_marker = ""
+    cachable = True
+
+    def __init__(self, path=None, fo=None, ref=None, **kwargs):
+        """
+
+        Parameters
+        ----------
+        path: str (optional)
+            Local location of the repo (uses current directory if not given).
+            May be deprecated in favour of ``fo``. When used with a higher
+            level function such as fsspec.open(), may be of the form
+            "git://[path-to-repo[:]][ref@]path/to/file" (but the actual
+            file path should not contain "@" or ":").
+        fo: str (optional)
+            Same as ``path``, but passed as part of a chained URL. This one
+            takes precedence if both are given.
+        ref: str (optional)
+            Reference to work with, could be a hash, tag or branch name. Defaults
+            to current working tree. Note that ``ls`` and ``open`` also take hash,
+            so this becomes the default for those operations
+        kwargs
+        """
+        super().__init__(**kwargs)
+        self.repo = pygit2.Repository(fo or path or os.getcwd())
+        self.ref = ref or "master"
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        path = super()._strip_protocol(path).lstrip("/")
+        if ":" in path:
+            path = path.split(":", 1)[1]
+        if "@" in path:
+            path = path.split("@", 1)[1]
+        return path.lstrip("/")
+
+    def _path_to_object(self, path, ref):
+        comm, ref = self.repo.resolve_refish(ref or self.ref)
+        parts = path.split("/")
+        tree = comm.tree
+        for part in parts:
+            if part and isinstance(tree, pygit2.Tree):
+                if part not in tree:
+                    raise FileNotFoundError(path)
+                tree = tree[part]
+        return tree
+
+    @staticmethod
+    def _get_kwargs_from_urls(path):
+        path = path.removeprefix("git://")
+        out = {}
+        if ":" in path:
+            out["path"], path = path.split(":", 1)
+        if "@" in path:
+            out["ref"], path = path.split("@", 1)
+        return out
+
+    @staticmethod
+    def _object_to_info(obj, path=None):
+        # obj.name and obj.filemode are None for the root tree!
+        is_dir = isinstance(obj, pygit2.Tree)
+        return {
+            "type": "directory" if is_dir else "file",
+            "name": (
+                "/".join([path, obj.name or ""]).lstrip("/") if path else obj.name
+            ),
+            "hex": str(obj.id),
+            "mode": "100644" if obj.filemode is None else f"{obj.filemode:o}",
+            "size": 0 if is_dir else obj.size,
+        }
+
+    def ls(self, path, detail=True, ref=None, **kwargs):
+        tree = self._path_to_object(self._strip_protocol(path), ref)
+        return [
+            GitFileSystem._object_to_info(obj, path)
+            if detail
+            else GitFileSystem._object_to_info(obj, path)["name"]
+            for obj in (tree if isinstance(tree, pygit2.Tree) else [tree])
+        ]
+
+    def info(self, path, ref=None, **kwargs):
+        tree = self._path_to_object(self._strip_protocol(path), ref)
+        return GitFileSystem._object_to_info(tree, path)
+
+    def ukey(self, path, ref=None):
+        return self.info(path, ref=ref)["hex"]
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        autocommit=True,
+        cache_options=None,
+        ref=None,
+        **kwargs,
+    ):
+        obj = self._path_to_object(path, ref or self.ref)
+        return MemoryFile(data=obj.data)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/github.py

@@ -0,0 +1,333 @@
+import base64
+import re
+
+import requests
+
+from ..spec import AbstractFileSystem
+from ..utils import infer_storage_options
+from .memory import MemoryFile
+
+
+class GithubFileSystem(AbstractFileSystem):
+    """Interface to files in github
+
+    An instance of this class provides the files residing within a remote github
+    repository. You may specify a point in the repos history, by SHA, branch
+    or tag (default is current master).
+
+    For files less than 1 MB in size, file content is returned directly in a
+    MemoryFile. For larger files, or for files tracked by git-lfs, file content
+    is returned as an HTTPFile wrapping the ``download_url`` provided by the
+    GitHub API.
+
+    When using fsspec.open, allows URIs of the form:
+
+    - "github://path/file", in which case you must specify org, repo and
+      may specify sha in the extra args
+    - 'github://org:repo@/precip/catalog.yml', where the org and repo are
+      part of the URI
+    - 'github://org:repo@sha/precip/catalog.yml', where the sha is also included
+
+    ``sha`` can be the full or abbreviated hex of the commit you want to fetch
+    from, or a branch or tag name (so long as it doesn't contain special characters
+    like "/", "?", which would have to be HTTP-encoded).
+
+    For authorised access, you must provide username and token, which can be made
+    at https://github.com/settings/tokens
+    """
+
+    url = "https://api.github.com/repos/{org}/{repo}/git/trees/{sha}"
+    content_url = "https://api.github.com/repos/{org}/{repo}/contents/{path}?ref={sha}"
+    protocol = "github"
+    timeout = (60, 60)  # connect, read timeouts
+
+    def __init__(
+        self, org, repo, sha=None, username=None, token=None, timeout=None, **kwargs
+    ):
+        super().__init__(**kwargs)
+        self.org = org
+        self.repo = repo
+        if (username is None) ^ (token is None):
+            raise ValueError("Auth required both username and token")
+        self.username = username
+        self.token = token
+        if timeout is not None:
+            self.timeout = timeout
+        if sha is None:
+            # look up default branch (not necessarily "master")
+            u = "https://api.github.com/repos/{org}/{repo}"
+            r = requests.get(
+                u.format(org=org, repo=repo), timeout=self.timeout, **self.kw
+            )
+            r.raise_for_status()
+            sha = r.json()["default_branch"]
+
+        self.root = sha
+        self.ls("")
+        try:
+            from .http import HTTPFileSystem
+
+            self.http_fs = HTTPFileSystem(**kwargs)
+        except ImportError:
+            self.http_fs = None
+
+    @property
+    def kw(self):
+        if self.username:
+            return {"auth": (self.username, self.token)}
+        return {}
+
+    @classmethod
+    def repos(cls, org_or_user, is_org=True):
+        """List repo names for given org or user
+
+        This may become the top level of the FS
+
+        Parameters
+        ----------
+        org_or_user: str
+            Name of the github org or user to query
+        is_org: bool (default True)
+            Whether the name is an organisation (True) or user (False)
+
+        Returns
+        -------
+        List of string
+        """
+        r = requests.get(
+            f"https://api.github.com/{['users', 'orgs'][is_org]}/{org_or_user}/repos",
+            timeout=cls.timeout,
+        )
+        r.raise_for_status()
+        return [repo["name"] for repo in r.json()]
+
+    @property
+    def tags(self):
+        """Names of tags in the repo"""
+        r = requests.get(
+            f"https://api.github.com/repos/{self.org}/{self.repo}/tags",
+            timeout=self.timeout,
+            **self.kw,
+        )
+        r.raise_for_status()
+        return [t["name"] for t in r.json()]
+
+    @property
+    def branches(self):
+        """Names of branches in the repo"""
+        r = requests.get(
+            f"https://api.github.com/repos/{self.org}/{self.repo}/branches",
+            timeout=self.timeout,
+            **self.kw,
+        )
+        r.raise_for_status()
+        return [t["name"] for t in r.json()]
+
+    @property
+    def refs(self):
+        """Named references, tags and branches"""
+        return {"tags": self.tags, "branches": self.branches}
+
+    def ls(self, path, detail=False, sha=None, _sha=None, **kwargs):
+        """List files at given path
+
+        Parameters
+        ----------
+        path: str
+            Location to list, relative to repo root
+        detail: bool
+            If True, returns list of dicts, one per file; if False, returns
+            list of full filenames only
+        sha: str (optional)
+            List at the given point in the repo history, branch or tag name or commit
+            SHA
+        _sha: str (optional)
+            List this specific tree object (used internally to descend into trees)
+        """
+        path = self._strip_protocol(path)
+        if path == "":
+            _sha = sha or self.root
+        if _sha is None:
+            parts = path.rstrip("/").split("/")
+            so_far = ""
+            _sha = sha or self.root
+            for part in parts:
+                out = self.ls(so_far, True, sha=sha, _sha=_sha)
+                so_far += "/" + part if so_far else part
+                out = [o for o in out if o["name"] == so_far]
+                if not out:
+                    raise FileNotFoundError(path)
+                out = out[0]
+                if out["type"] == "file":
+                    if detail:
+                        return [out]
+                    else:
+                        return path
+                _sha = out["sha"]
+        if path not in self.dircache or sha not in [self.root, None]:
+            r = requests.get(
+                self.url.format(org=self.org, repo=self.repo, sha=_sha),
+                timeout=self.timeout,
+                **self.kw,
+            )
+            if r.status_code == 404:
+                raise FileNotFoundError(path)
+            r.raise_for_status()
+            types = {"blob": "file", "tree": "directory"}
+            out = [
+                {
+                    "name": path + "/" + f["path"] if path else f["path"],
+                    "mode": f["mode"],
+                    "type": types[f["type"]],
+                    "size": f.get("size", 0),
+                    "sha": f["sha"],
+                }
+                for f in r.json()["tree"]
+                if f["type"] in types
+            ]
+            if sha in [self.root, None]:
+                self.dircache[path] = out
+        else:
+            out = self.dircache[path]
+        if detail:
+            return out
+        else:
+            return sorted([f["name"] for f in out])
+
+    def invalidate_cache(self, path=None):
+        self.dircache.clear()
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        opts = infer_storage_options(path)
+        if "username" not in opts:
+            return super()._strip_protocol(path)
+        return opts["path"].lstrip("/")
+
+    @staticmethod
+    def _get_kwargs_from_urls(path):
+        opts = infer_storage_options(path)
+        if "username" not in opts:
+            return {}
+        out = {"org": opts["username"], "repo": opts["password"]}
+        if opts["host"]:
+            out["sha"] = opts["host"]
+        return out
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        cache_options=None,
+        sha=None,
+        **kwargs,
+    ):
+        if mode != "rb":
+            raise NotImplementedError
+
+        # construct a url to hit the GitHub API's repo contents API
+        url = self.content_url.format(
+            org=self.org, repo=self.repo, path=path, sha=sha or self.root
+        )
+
+        # make a request to this API, and parse the response as JSON
+        r = requests.get(url, timeout=self.timeout, **self.kw)
+        if r.status_code == 404:
+            raise FileNotFoundError(path)
+        r.raise_for_status()
+        content_json = r.json()
+
+        # if the response's content key is not empty, try to parse it as base64
+        if content_json["content"]:
+            content = base64.b64decode(content_json["content"])
+
+            # as long as the content does not start with the string
+            # "version https://git-lfs.github.com/"
+            # then it is probably not a git-lfs pointer and we can just return
+            # the content directly
+            if not content.startswith(b"version https://git-lfs.github.com/"):
+                return MemoryFile(None, None, content)
+
+        # we land here if the content was not present in the first response
+        # (regular file over 1MB or git-lfs tracked file)
+        # in this case, we get let the HTTPFileSystem handle the download
+        if self.http_fs is None:
+            raise ImportError(
+                "Please install fsspec[http] to access github files >1 MB "
+                "or git-lfs tracked files."
+            )
+        return self.http_fs.open(
+            content_json["download_url"],
+            mode=mode,
+            block_size=block_size,
+            cache_options=cache_options,
+            **kwargs,
+        )
+
+    def rm(self, path, recursive=False, maxdepth=None, message=None):
+        path = self.expand_path(path, recursive=recursive, maxdepth=maxdepth)
+        for p in reversed(path):
+            self.rm_file(p, message=message)
+
+    def rm_file(self, path, message=None, **kwargs):
+        """
+        Remove a file from a specified branch using a given commit message.
+
+        Since Github DELETE operation requires a branch name, and we can't reliably
+        determine whether the provided SHA refers to a branch, tag, or commit, we
+        assume it's a branch. If it's not, the user will encounter an error when
+        attempting to retrieve the file SHA or delete the file.
+
+        Parameters
+        ----------
+        path: str
+            The file's location relative to the repository root.
+        message: str, optional
+            The commit message for the deletion.
+        """
+
+        if not self.username:
+            raise ValueError("Authentication required")
+
+        path = self._strip_protocol(path)
+
+        # Attempt to get SHA from cache or Github API
+        sha = self._get_sha_from_cache(path)
+        if not sha:
+            url = self.content_url.format(
+                org=self.org, repo=self.repo, path=path.lstrip("/"), sha=self.root
+            )
+            r = requests.get(url, timeout=self.timeout, **self.kw)
+            if r.status_code == 404:
+                raise FileNotFoundError(path)
+            r.raise_for_status()
+            sha = r.json()["sha"]
+
+        # Delete the file
+        delete_url = self.content_url.format(
+            org=self.org, repo=self.repo, path=path, sha=self.root
+        )
+        branch = self.root
+        data = {
+            "message": message or f"Delete {path}",
+            "sha": sha,
+            **({"branch": branch} if branch else {}),
+        }
+
+        r = requests.delete(delete_url, json=data, timeout=self.timeout, **self.kw)
+        error_message = r.json().get("message", "")
+        if re.search(r"Branch .+ not found", error_message):
+            error = "Remove only works when the filesystem is initialised from a branch or default (None)"
+            raise ValueError(error)
+        r.raise_for_status()
+
+        self.invalidate_cache(path)
+
+    def _get_sha_from_cache(self, path):
+        for entries in self.dircache.values():
+            for entry in entries:
+                entry_path = entry.get("name")
+                if entry_path and entry_path == path and "sha" in entry:
+                    return entry["sha"]
+        return None

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/http.py

@@ -0,0 +1,891 @@
+import asyncio
+import io
+import logging
+import re
+import weakref
+from copy import copy
+from urllib.parse import urlparse
+
+import aiohttp
+import yarl
+
+from fsspec.asyn import AbstractAsyncStreamedFile, AsyncFileSystem, sync, sync_wrapper
+from fsspec.callbacks import DEFAULT_CALLBACK
+from fsspec.exceptions import FSTimeoutError
+from fsspec.spec import AbstractBufferedFile
+from fsspec.utils import (
+    DEFAULT_BLOCK_SIZE,
+    glob_translate,
+    isfilelike,
+    nullcontext,
+    tokenize,
+)
+
+from ..caching import AllBytes
+
+# https://stackoverflow.com/a/15926317/3821154
+ex = re.compile(r"""<(a|A)\s+(?:[^>]*?\s+)?(href|HREF)=["'](?P<url>[^"']+)""")
+ex2 = re.compile(r"""(?P<url>http[s]?://[-a-zA-Z0-9@:%_+.~#?&/=]+)""")
+logger = logging.getLogger("fsspec.http")
+
+
+async def get_client(**kwargs):
+    return aiohttp.ClientSession(**kwargs)
+
+
+class HTTPFileSystem(AsyncFileSystem):
+    """
+    Simple File-System for fetching data via HTTP(S)
+
+    ``ls()`` is implemented by loading the parent page and doing a regex
+    match on the result. If simple_link=True, anything of the form
+    "http(s)://server.com/stuff?thing=other"; otherwise only links within
+    HTML href tags will be used.
+    """
+
+    protocol = ("http", "https")
+    sep = "/"
+
+    def __init__(
+        self,
+        simple_links=True,
+        block_size=None,
+        same_scheme=True,
+        size_policy=None,
+        cache_type="bytes",
+        cache_options=None,
+        asynchronous=False,
+        loop=None,
+        client_kwargs=None,
+        get_client=get_client,
+        encoded=False,
+        **storage_options,
+    ):
+        """
+        NB: if this is called async, you must await set_client
+
+        Parameters
+        ----------
+        block_size: int
+            Blocks to read bytes; if 0, will default to raw requests file-like
+            objects instead of HTTPFile instances
+        simple_links: bool
+            If True, will consider both HTML <a> tags and anything that looks
+            like a URL; if False, will consider only the former.
+        same_scheme: True
+            When doing ls/glob, if this is True, only consider paths that have
+            http/https matching the input URLs.
+        size_policy: this argument is deprecated
+        client_kwargs: dict
+            Passed to aiohttp.ClientSession, see
+            https://docs.aiohttp.org/en/stable/client_reference.html
+            For example, ``{'auth': aiohttp.BasicAuth('user', 'pass')}``
+        get_client: Callable[..., aiohttp.ClientSession]
+            A callable, which takes keyword arguments and constructs
+            an aiohttp.ClientSession. Its state will be managed by
+            the HTTPFileSystem class.
+        storage_options: key-value
+            Any other parameters passed on to requests
+        cache_type, cache_options: defaults used in open()
+        """
+        super().__init__(self, asynchronous=asynchronous, loop=loop, **storage_options)
+        self.block_size = block_size if block_size is not None else DEFAULT_BLOCK_SIZE
+        self.simple_links = simple_links
+        self.same_schema = same_scheme
+        self.cache_type = cache_type
+        self.cache_options = cache_options
+        self.client_kwargs = client_kwargs or {}
+        self.get_client = get_client
+        self.encoded = encoded
+        self.kwargs = storage_options
+        self._session = None
+
+        # Clean caching-related parameters from `storage_options`
+        # before propagating them as `request_options` through `self.kwargs`.
+        # TODO: Maybe rename `self.kwargs` to `self.request_options` to make
+        #       it clearer.
+        request_options = copy(storage_options)
+        self.use_listings_cache = request_options.pop("use_listings_cache", False)
+        request_options.pop("listings_expiry_time", None)
+        request_options.pop("max_paths", None)
+        request_options.pop("skip_instance_cache", None)
+        self.kwargs = request_options
+
+    @property
+    def fsid(self):
+        return "http"
+
+    def encode_url(self, url):
+        return yarl.URL(url, encoded=self.encoded)
+
+    @staticmethod
+    def close_session(loop, session):
+        if loop is not None and loop.is_running():
+            try:
+                sync(loop, session.close, timeout=0.1)
+                return
+            except (TimeoutError, FSTimeoutError, NotImplementedError):
+                pass
+        connector = getattr(session, "_connector", None)
+        if connector is not None:
+            # close after loop is dead
+            connector._close()
+
+    async def set_session(self):
+        if self._session is None:
+            self._session = await self.get_client(loop=self.loop, **self.client_kwargs)
+            if not self.asynchronous:
+                weakref.finalize(self, self.close_session, self.loop, self._session)
+        return self._session
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        """For HTTP, we always want to keep the full URL"""
+        return path
+
+    @classmethod
+    def _parent(cls, path):
+        # override, since _strip_protocol is different for URLs
+        par = super()._parent(path)
+        if len(par) > 7:  # "http://..."
+            return par
+        return ""
+
+    async def _ls_real(self, url, detail=True, **kwargs):
+        # ignoring URL-encoded arguments
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        logger.debug(url)
+        session = await self.set_session()
+        async with session.get(self.encode_url(url), **self.kwargs) as r:
+            self._raise_not_found_for_status(r, url)
+
+            if "Content-Type" in r.headers:
+                mimetype = r.headers["Content-Type"].partition(";")[0]
+            else:
+                mimetype = None
+
+            if mimetype in ("text/html", None):
+                try:
+                    text = await r.text(errors="ignore")
+                    if self.simple_links:
+                        links = ex2.findall(text) + [u[2] for u in ex.findall(text)]
+                    else:
+                        links = [u[2] for u in ex.findall(text)]
+                except UnicodeDecodeError:
+                    links = []  # binary, not HTML
+            else:
+                links = []
+
+        out = set()
+        parts = urlparse(url)
+        for l in links:
+            if isinstance(l, tuple):
+                l = l[1]
+            if l.startswith("/") and len(l) > 1:
+                # absolute URL on this server
+                l = f"{parts.scheme}://{parts.netloc}{l}"
+            if l.startswith("http"):
+                if self.same_schema and l.startswith(url.rstrip("/") + "/"):
+                    out.add(l)
+                elif l.replace("https", "http").startswith(
+                    url.replace("https", "http").rstrip("/") + "/"
+                ):
+                    # allowed to cross http <-> https
+                    out.add(l)
+            else:
+                if l not in ["..", "../"]:
+                    # Ignore FTP-like "parent"
+                    out.add("/".join([url.rstrip("/"), l.lstrip("/")]))
+        if not out and url.endswith("/"):
+            out = await self._ls_real(url.rstrip("/"), detail=False)
+        if detail:
+            return [
+                {
+                    "name": u,
+                    "size": None,
+                    "type": "directory" if u.endswith("/") else "file",
+                }
+                for u in out
+            ]
+        else:
+            return sorted(out)
+
+    async def _ls(self, url, detail=True, **kwargs):
+        if self.use_listings_cache and url in self.dircache:
+            out = self.dircache[url]
+        else:
+            out = await self._ls_real(url, detail=detail, **kwargs)
+            self.dircache[url] = out
+        return out
+
+    ls = sync_wrapper(_ls)
+
+    def _raise_not_found_for_status(self, response, url):
+        """
+        Raises FileNotFoundError for 404s, otherwise uses raise_for_status.
+        """
+        if response.status == 404:
+            raise FileNotFoundError(url)
+        response.raise_for_status()
+
+    async def _cat_file(self, url, start=None, end=None, **kwargs):
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        logger.debug(url)
+
+        if start is not None or end is not None:
+            if start == end:
+                return b""
+            headers = kw.pop("headers", {}).copy()
+
+            headers["Range"] = await self._process_limits(url, start, end)
+            kw["headers"] = headers
+        session = await self.set_session()
+        async with session.get(self.encode_url(url), **kw) as r:
+            out = await r.read()
+            self._raise_not_found_for_status(r, url)
+        return out
+
+    async def _get_file(
+        self, rpath, lpath, chunk_size=5 * 2**20, callback=DEFAULT_CALLBACK, **kwargs
+    ):
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        logger.debug(rpath)
+        session = await self.set_session()
+        async with session.get(self.encode_url(rpath), **kw) as r:
+            try:
+                size = int(r.headers["content-length"])
+            except (ValueError, KeyError):
+                size = None
+
+            callback.set_size(size)
+            self._raise_not_found_for_status(r, rpath)
+            if isfilelike(lpath):
+                outfile = lpath
+            else:
+                outfile = open(lpath, "wb")  # noqa: ASYNC230
+
+            try:
+                chunk = True
+                while chunk:
+                    chunk = await r.content.read(chunk_size)
+                    outfile.write(chunk)
+                    callback.relative_update(len(chunk))
+            finally:
+                if not isfilelike(lpath):
+                    outfile.close()
+
+    async def _put_file(
+        self,
+        lpath,
+        rpath,
+        chunk_size=5 * 2**20,
+        callback=DEFAULT_CALLBACK,
+        method="post",
+        mode="overwrite",
+        **kwargs,
+    ):
+        if mode != "overwrite":
+            raise NotImplementedError("Exclusive write")
+
+        async def gen_chunks():
+            # Support passing arbitrary file-like objects
+            # and use them instead of streams.
+            if isinstance(lpath, io.IOBase):
+                context = nullcontext(lpath)
+                use_seek = False  # might not support seeking
+            else:
+                context = open(lpath, "rb")  # noqa: ASYNC230
+                use_seek = True
+
+            with context as f:
+                if use_seek:
+                    callback.set_size(f.seek(0, 2))
+                    f.seek(0)
+                else:
+                    callback.set_size(getattr(f, "size", None))
+
+                chunk = f.read(chunk_size)
+                while chunk:
+                    yield chunk
+                    callback.relative_update(len(chunk))
+                    chunk = f.read(chunk_size)
+
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        session = await self.set_session()
+
+        method = method.lower()
+        if method not in ("post", "put"):
+            raise ValueError(
+                f"method has to be either 'post' or 'put', not: {method!r}"
+            )
+
+        meth = getattr(session, method)
+        async with meth(self.encode_url(rpath), data=gen_chunks(), **kw) as resp:
+            self._raise_not_found_for_status(resp, rpath)
+
+    async def _exists(self, path, **kwargs):
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        try:
+            logger.debug(path)
+            session = await self.set_session()
+            r = await session.get(self.encode_url(path), **kw)
+            async with r:
+                return r.status < 400
+        except aiohttp.ClientError:
+            return False
+
+    async def _isfile(self, path, **kwargs):
+        return await self._exists(path, **kwargs)
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        autocommit=None,  # XXX: This differs from the base class.
+        cache_type=None,
+        cache_options=None,
+        size=None,
+        **kwargs,
+    ):
+        """Make a file-like object
+
+        Parameters
+        ----------
+        path: str
+            Full URL with protocol
+        mode: string
+            must be "rb"
+        block_size: int or None
+            Bytes to download in one request; use instance value if None. If
+            zero, will return a streaming Requests file-like instance.
+        kwargs: key-value
+            Any other parameters, passed to requests calls
+        """
+        if mode != "rb":
+            raise NotImplementedError
+        block_size = block_size if block_size is not None else self.block_size
+        kw = self.kwargs.copy()
+        kw["asynchronous"] = self.asynchronous
+        kw.update(kwargs)
+        info = {}
+        size = size or info.update(self.info(path, **kwargs)) or info["size"]
+        session = sync(self.loop, self.set_session)
+        if block_size and size and info.get("partial", True):
+            return HTTPFile(
+                self,
+                path,
+                session=session,
+                block_size=block_size,
+                mode=mode,
+                size=size,
+                cache_type=cache_type or self.cache_type,
+                cache_options=cache_options or self.cache_options,
+                loop=self.loop,
+                **kw,
+            )
+        else:
+            return HTTPStreamFile(
+                self,
+                path,
+                mode=mode,
+                loop=self.loop,
+                session=session,
+                **kw,
+            )
+
+    async def open_async(self, path, mode="rb", size=None, **kwargs):
+        session = await self.set_session()
+        if size is None:
+            try:
+                size = (await self._info(path, **kwargs))["size"]
+            except FileNotFoundError:
+                pass
+        return AsyncStreamFile(
+            self,
+            path,
+            loop=self.loop,
+            session=session,
+            size=size,
+            **kwargs,
+        )
+
+    def ukey(self, url):
+        """Unique identifier; assume HTTP files are static, unchanging"""
+        return tokenize(url, self.kwargs, self.protocol)
+
+    async def _info(self, url, **kwargs):
+        """Get info of URL
+
+        Tries to access location via HEAD, and then GET methods, but does
+        not fetch the data.
+
+        It is possible that the server does not supply any size information, in
+        which case size will be given as None (and certain operations on the
+        corresponding file will not work).
+        """
+        info = {}
+        session = await self.set_session()
+
+        for policy in ["head", "get"]:
+            try:
+                info.update(
+                    await _file_info(
+                        self.encode_url(url),
+                        size_policy=policy,
+                        session=session,
+                        **self.kwargs,
+                        **kwargs,
+                    )
+                )
+                if info.get("size") is not None:
+                    break
+            except Exception as exc:
+                if policy == "get":
+                    # If get failed, then raise a FileNotFoundError
+                    raise FileNotFoundError(url) from exc
+                logger.debug("", exc_info=exc)
+
+        return {"name": url, "size": None, **info, "type": "file"}
+
+    async def _glob(self, path, maxdepth=None, **kwargs):
+        """
+        Find files by glob-matching.
+
+        This implementation is idntical to the one in AbstractFileSystem,
+        but "?" is not considered as a character for globbing, because it is
+        so common in URLs, often identifying the "query" part.
+        """
+        if maxdepth is not None and maxdepth < 1:
+            raise ValueError("maxdepth must be at least 1")
+        import re
+
+        ends_with_slash = path.endswith("/")  # _strip_protocol strips trailing slash
+        path = self._strip_protocol(path)
+        append_slash_to_dirname = ends_with_slash or path.endswith(("/**", "/*"))
+        idx_star = path.find("*") if path.find("*") >= 0 else len(path)
+        idx_brace = path.find("[") if path.find("[") >= 0 else len(path)
+
+        min_idx = min(idx_star, idx_brace)
+
+        detail = kwargs.pop("detail", False)
+
+        if not has_magic(path):
+            if await self._exists(path, **kwargs):
+                if not detail:
+                    return [path]
+                else:
+                    return {path: await self._info(path, **kwargs)}
+            else:
+                if not detail:
+                    return []  # glob of non-existent returns empty
+                else:
+                    return {}
+        elif "/" in path[:min_idx]:
+            min_idx = path[:min_idx].rindex("/")
+            root = path[: min_idx + 1]
+            depth = path[min_idx + 1 :].count("/") + 1
+        else:
+            root = ""
+            depth = path[min_idx + 1 :].count("/") + 1
+
+        if "**" in path:
+            if maxdepth is not None:
+                idx_double_stars = path.find("**")
+                depth_double_stars = path[idx_double_stars:].count("/") + 1
+                depth = depth - depth_double_stars + maxdepth
+            else:
+                depth = None
+
+        allpaths = await self._find(
+            root, maxdepth=depth, withdirs=True, detail=True, **kwargs
+        )
+
+        pattern = glob_translate(path + ("/" if ends_with_slash else ""))
+        pattern = re.compile(pattern)
+
+        out = {
+            (
+                p.rstrip("/")
+                if not append_slash_to_dirname
+                and info["type"] == "directory"
+                and p.endswith("/")
+                else p
+            ): info
+            for p, info in sorted(allpaths.items())
+            if pattern.match(p.rstrip("/"))
+        }
+
+        if detail:
+            return out
+        else:
+            return list(out)
+
+    async def _isdir(self, path):
+        # override, since all URLs are (also) files
+        try:
+            return bool(await self._ls(path))
+        except (FileNotFoundError, ValueError):
+            return False
+
+    async def _pipe_file(self, path, value, mode="overwrite", **kwargs):
+        """
+        Write bytes to a remote file over HTTP.
+
+        Parameters
+        ----------
+        path : str
+            Target URL where the data should be written
+        value : bytes
+            Data to be written
+        mode : str
+            How to write to the file - 'overwrite' or 'append'
+        **kwargs : dict
+            Additional parameters to pass to the HTTP request
+        """
+        url = self._strip_protocol(path)
+        headers = kwargs.pop("headers", {})
+        headers["Content-Length"] = str(len(value))
+
+        session = await self.set_session()
+
+        async with session.put(url, data=value, headers=headers, **kwargs) as r:
+            r.raise_for_status()
+
+
+class HTTPFile(AbstractBufferedFile):
+    """
+    A file-like object pointing to a remote HTTP(S) resource
+
+    Supports only reading, with read-ahead of a predetermined block-size.
+
+    In the case that the server does not supply the filesize, only reading of
+    the complete file in one go is supported.
+
+    Parameters
+    ----------
+    url: str
+        Full URL of the remote resource, including the protocol
+    session: aiohttp.ClientSession or None
+        All calls will be made within this session, to avoid restarting
+        connections where the server allows this
+    block_size: int or None
+        The amount of read-ahead to do, in bytes. Default is 5MB, or the value
+        configured for the FileSystem creating this file
+    size: None or int
+        If given, this is the size of the file in bytes, and we don't attempt
+        to call the server to find the value.
+    kwargs: all other key-values are passed to requests calls.
+    """
+
+    def __init__(
+        self,
+        fs,
+        url,
+        session=None,
+        block_size=None,
+        mode="rb",
+        cache_type="bytes",
+        cache_options=None,
+        size=None,
+        loop=None,
+        asynchronous=False,
+        **kwargs,
+    ):
+        if mode != "rb":
+            raise NotImplementedError("File mode not supported")
+        self.asynchronous = asynchronous
+        self.loop = loop
+        self.url = url
+        self.session = session
+        self.details = {"name": url, "size": size, "type": "file"}
+        super().__init__(
+            fs=fs,
+            path=url,
+            mode=mode,
+            block_size=block_size,
+            cache_type=cache_type,
+            cache_options=cache_options,
+            **kwargs,
+        )
+
+    def read(self, length=-1):
+        """Read bytes from file
+
+        Parameters
+        ----------
+        length: int
+            Read up to this many bytes. If negative, read all content to end of
+            file. If the server has not supplied the filesize, attempting to
+            read only part of the data will raise a ValueError.
+        """
+        if (
+            (length < 0 and self.loc == 0)  # explicit read all
+            # but not when the size is known and fits into a block anyways
+            and not (self.size is not None and self.size <= self.blocksize)
+        ):
+            self._fetch_all()
+        if self.size is None:
+            if length < 0:
+                self._fetch_all()
+        else:
+            length = min(self.size - self.loc, length)
+        return super().read(length)
+
+    async def async_fetch_all(self):
+        """Read whole file in one shot, without caching
+
+        This is only called when position is still at zero,
+        and read() is called without a byte-count.
+        """
+        logger.debug(f"Fetch all for {self}")
+        if not isinstance(self.cache, AllBytes):
+            r = await self.session.get(self.fs.encode_url(self.url), **self.kwargs)
+            async with r:
+                r.raise_for_status()
+                out = await r.read()
+                self.cache = AllBytes(
+                    size=len(out), fetcher=None, blocksize=None, data=out
+                )
+                self.size = len(out)
+
+    _fetch_all = sync_wrapper(async_fetch_all)
+
+    def _parse_content_range(self, headers):
+        """Parse the Content-Range header"""
+        s = headers.get("Content-Range", "")
+        m = re.match(r"bytes (\d+-\d+|\*)/(\d+|\*)", s)
+        if not m:
+            return None, None, None
+
+        if m[1] == "*":
+            start = end = None
+        else:
+            start, end = [int(x) for x in m[1].split("-")]
+        total = None if m[2] == "*" else int(m[2])
+        return start, end, total
+
+    async def async_fetch_range(self, start, end):
+        """Download a block of data
+
+        The expectation is that the server returns only the requested bytes,
+        with HTTP code 206. If this is not the case, we first check the headers,
+        and then stream the output - if the data size is bigger than we
+        requested, an exception is raised.
+        """
+        logger.debug(f"Fetch range for {self}: {start}-{end}")
+        kwargs = self.kwargs.copy()
+        headers = kwargs.pop("headers", {}).copy()
+        headers["Range"] = f"bytes={start}-{end - 1}"
+        logger.debug(f"{self.url} : {headers['Range']}")
+        r = await self.session.get(
+            self.fs.encode_url(self.url), headers=headers, **kwargs
+        )
+        async with r:
+            if r.status == 416:
+                # range request outside file
+                return b""
+            r.raise_for_status()
+
+            # If the server has handled the range request, it should reply
+            # with status 206 (partial content). But we'll guess that a suitable
+            # Content-Range header or a Content-Length no more than the
+            # requested range also mean we have got the desired range.
+            response_is_range = (
+                r.status == 206
+                or self._parse_content_range(r.headers)[0] == start
+                or int(r.headers.get("Content-Length", end + 1)) <= end - start
+            )
+
+            if response_is_range:
+                # partial content, as expected
+                out = await r.read()
+            elif start > 0:
+                raise ValueError(
+                    "The HTTP server doesn't appear to support range requests. "
+                    "Only reading this file from the beginning is supported. "
+                    "Open with block_size=0 for a streaming file interface."
+                )
+            else:
+                # Response is not a range, but we want the start of the file,
+                # so we can read the required amount anyway.
+                cl = 0
+                out = []
+                while True:
+                    chunk = await r.content.read(2**20)
+                    # data size unknown, let's read until we have enough
+                    if chunk:
+                        out.append(chunk)
+                        cl += len(chunk)
+                        if cl > end - start:
+                            break
+                    else:
+                        break
+                out = b"".join(out)[: end - start]
+            return out
+
+    _fetch_range = sync_wrapper(async_fetch_range)
+
+
+magic_check = re.compile("([*[])")
+
+
+def has_magic(s):
+    match = magic_check.search(s)
+    return match is not None
+
+
+class HTTPStreamFile(AbstractBufferedFile):
+    def __init__(self, fs, url, mode="rb", loop=None, session=None, **kwargs):
+        self.asynchronous = kwargs.pop("asynchronous", False)
+        self.url = url
+        self.loop = loop
+        self.session = session
+        if mode != "rb":
+            raise ValueError
+        self.details = {"name": url, "size": None}
+        super().__init__(fs=fs, path=url, mode=mode, cache_type="none", **kwargs)
+
+        async def cor():
+            r = await self.session.get(self.fs.encode_url(url), **kwargs).__aenter__()
+            self.fs._raise_not_found_for_status(r, url)
+            return r
+
+        self.r = sync(self.loop, cor)
+        self.loop = fs.loop
+
+    def seek(self, loc, whence=0):
+        if loc == 0 and whence == 1:
+            return
+        if loc == self.loc and whence == 0:
+            return
+        raise ValueError("Cannot seek streaming HTTP file")
+
+    async def _read(self, num=-1):
+        out = await self.r.content.read(num)
+        self.loc += len(out)
+        return out
+
+    read = sync_wrapper(_read)
+
+    async def _close(self):
+        self.r.close()
+
+    def close(self):
+        asyncio.run_coroutine_threadsafe(self._close(), self.loop)
+        super().close()
+
+
+class AsyncStreamFile(AbstractAsyncStreamedFile):
+    def __init__(
+        self, fs, url, mode="rb", loop=None, session=None, size=None, **kwargs
+    ):
+        self.url = url
+        self.session = session
+        self.r = None
+        if mode != "rb":
+            raise ValueError
+        self.details = {"name": url, "size": None}
+        self.kwargs = kwargs
+        super().__init__(fs=fs, path=url, mode=mode, cache_type="none")
+        self.size = size
+
+    async def read(self, num=-1):
+        if self.r is None:
+            r = await self.session.get(
+                self.fs.encode_url(self.url), **self.kwargs
+            ).__aenter__()
+            self.fs._raise_not_found_for_status(r, self.url)
+            self.r = r
+        out = await self.r.content.read(num)
+        self.loc += len(out)
+        return out
+
+    async def close(self):
+        if self.r is not None:
+            self.r.close()
+            self.r = None
+        await super().close()
+
+
+async def get_range(session, url, start, end, file=None, **kwargs):
+    # explicit get a range when we know it must be safe
+    kwargs = kwargs.copy()
+    headers = kwargs.pop("headers", {}).copy()
+    headers["Range"] = f"bytes={start}-{end - 1}"
+    r = await session.get(url, headers=headers, **kwargs)
+    r.raise_for_status()
+    async with r:
+        out = await r.read()
+    if file:
+        with open(file, "r+b") as f:  # noqa: ASYNC230
+            f.seek(start)
+            f.write(out)
+    else:
+        return out
+
+
+async def _file_info(url, session, size_policy="head", **kwargs):
+    """Call HEAD on the server to get details about the file (size/checksum etc.)
+
+    Default operation is to explicitly allow redirects and use encoding
+    'identity' (no compression) to get the true size of the target.
+    """
+    logger.debug("Retrieve file size for %s", url)
+    kwargs = kwargs.copy()
+    ar = kwargs.pop("allow_redirects", True)
+    head = kwargs.get("headers", {}).copy()
+    head["Accept-Encoding"] = "identity"
+    kwargs["headers"] = head
+
+    info = {}
+    if size_policy == "head":
+        r = await session.head(url, allow_redirects=ar, **kwargs)
+    elif size_policy == "get":
+        r = await session.get(url, allow_redirects=ar, **kwargs)
+    else:
+        raise TypeError(f'size_policy must be "head" or "get", got {size_policy}')
+    async with r:
+        r.raise_for_status()
+
+        if "Content-Length" in r.headers:
+            # Some servers may choose to ignore Accept-Encoding and return
+            # compressed content, in which case the returned size is unreliable.
+            if "Content-Encoding" not in r.headers or r.headers["Content-Encoding"] in [
+                "identity",
+                "",
+            ]:
+                info["size"] = int(r.headers["Content-Length"])
+        elif "Content-Range" in r.headers:
+            info["size"] = int(r.headers["Content-Range"].split("/")[1])
+
+        if "Content-Type" in r.headers:
+            info["mimetype"] = r.headers["Content-Type"].partition(";")[0]
+
+        if r.headers.get("Accept-Ranges") == "none":
+            # Some servers may explicitly discourage partial content requests, but
+            # the lack of "Accept-Ranges" does not always indicate they would fail
+            info["partial"] = False
+
+        info["url"] = str(r.url)
+
+        for checksum_field in ["ETag", "Content-MD5", "Digest", "Last-Modified"]:
+            if r.headers.get(checksum_field):
+                info[checksum_field] = r.headers[checksum_field]
+
+    return info
+
+
+async def _file_size(url, session=None, *args, **kwargs):
+    if session is None:
+        session = await get_client()
+    info = await _file_info(url, session=session, *args, **kwargs)
+    return info.get("size")
+
+
+file_size = sync_wrapper(_file_size)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/http_sync.py

@@ -0,0 +1,931 @@
+"""This file is largely copied from http.py"""
+
+import io
+import logging
+import re
+import urllib.error
+import urllib.parse
+from copy import copy
+from json import dumps, loads
+from urllib.parse import urlparse
+
+try:
+    import yarl
+except (ImportError, ModuleNotFoundError, OSError):
+    yarl = False
+
+from fsspec.callbacks import _DEFAULT_CALLBACK
+from fsspec.registry import register_implementation
+from fsspec.spec import AbstractBufferedFile, AbstractFileSystem
+from fsspec.utils import DEFAULT_BLOCK_SIZE, isfilelike, nullcontext, tokenize
+
+from ..caching import AllBytes
+
+# https://stackoverflow.com/a/15926317/3821154
+ex = re.compile(r"""<(a|A)\s+(?:[^>]*?\s+)?(href|HREF)=["'](?P<url>[^"']+)""")
+ex2 = re.compile(r"""(?P<url>http[s]?://[-a-zA-Z0-9@:%_+.~#?&/=]+)""")
+logger = logging.getLogger("fsspec.http")
+
+
+class JsHttpException(urllib.error.HTTPError): ...
+
+
+class StreamIO(io.BytesIO):
+    # fake class, so you can set attributes on it
+    # will eventually actually stream
+    ...
+
+
+class ResponseProxy:
+    """Looks like a requests response"""
+
+    def __init__(self, req, stream=False):
+        self.request = req
+        self.stream = stream
+        self._data = None
+        self._headers = None
+
+    @property
+    def raw(self):
+        if self._data is None:
+            b = self.request.response.to_bytes()
+            if self.stream:
+                self._data = StreamIO(b)
+            else:
+                self._data = b
+        return self._data
+
+    def close(self):
+        if hasattr(self, "_data"):
+            del self._data
+
+    @property
+    def headers(self):
+        if self._headers is None:
+            self._headers = dict(
+                [
+                    _.split(": ")
+                    for _ in self.request.getAllResponseHeaders().strip().split("\r\n")
+                ]
+            )
+        return self._headers
+
+    @property
+    def status_code(self):
+        return int(self.request.status)
+
+    def raise_for_status(self):
+        if not self.ok:
+            raise JsHttpException(
+                self.url, self.status_code, self.reason, self.headers, None
+            )
+
+    def iter_content(self, chunksize, *_, **__):
+        while True:
+            out = self.raw.read(chunksize)
+            if out:
+                yield out
+            else:
+                break
+
+    @property
+    def reason(self):
+        return self.request.statusText
+
+    @property
+    def ok(self):
+        return self.status_code < 400
+
+    @property
+    def url(self):
+        return self.request.response.responseURL
+
+    @property
+    def text(self):
+        # TODO: encoding from headers
+        return self.content.decode()
+
+    @property
+    def content(self):
+        self.stream = False
+        return self.raw
+
+    def json(self):
+        return loads(self.text)
+
+
+class RequestsSessionShim:
+    def __init__(self):
+        self.headers = {}
+
+    def request(
+        self,
+        method,
+        url,
+        params=None,
+        data=None,
+        headers=None,
+        cookies=None,
+        files=None,
+        auth=None,
+        timeout=None,
+        allow_redirects=None,
+        proxies=None,
+        hooks=None,
+        stream=None,
+        verify=None,
+        cert=None,
+        json=None,
+    ):
+        from js import Blob, XMLHttpRequest
+
+        logger.debug("JS request: %s %s", method, url)
+
+        if cert or verify or proxies or files or cookies or hooks:
+            raise NotImplementedError
+        if data and json:
+            raise ValueError("Use json= or data=, not both")
+        req = XMLHttpRequest.new()
+        extra = auth if auth else ()
+        if params:
+            url = f"{url}?{urllib.parse.urlencode(params)}"
+        req.open(method, url, False, *extra)
+        if timeout:
+            req.timeout = timeout
+        if headers:
+            for k, v in headers.items():
+                req.setRequestHeader(k, v)
+
+        req.setRequestHeader("Accept", "application/octet-stream")
+        req.responseType = "arraybuffer"
+        if json:
+            blob = Blob.new([dumps(data)], {type: "application/json"})
+            req.send(blob)
+        elif data:
+            if isinstance(data, io.IOBase):
+                data = data.read()
+            blob = Blob.new([data], {type: "application/octet-stream"})
+            req.send(blob)
+        else:
+            req.send(None)
+        return ResponseProxy(req, stream=stream)
+
+    def get(self, url, **kwargs):
+        return self.request("GET", url, **kwargs)
+
+    def head(self, url, **kwargs):
+        return self.request("HEAD", url, **kwargs)
+
+    def post(self, url, **kwargs):
+        return self.request("POST}", url, **kwargs)
+
+    def put(self, url, **kwargs):
+        return self.request("PUT", url, **kwargs)
+
+    def patch(self, url, **kwargs):
+        return self.request("PATCH", url, **kwargs)
+
+    def delete(self, url, **kwargs):
+        return self.request("DELETE", url, **kwargs)
+
+
+class HTTPFileSystem(AbstractFileSystem):
+    """
+    Simple File-System for fetching data via HTTP(S)
+
+    This is the BLOCKING version of the normal HTTPFileSystem. It uses
+    requests in normal python and the JS runtime in pyodide.
+
+    ***This implementation is extremely experimental, do not use unless
+    you are testing pyodide/pyscript integration***
+    """
+
+    protocol = ("http", "https", "sync-http", "sync-https")
+    sep = "/"
+
+    def __init__(
+        self,
+        simple_links=True,
+        block_size=None,
+        same_scheme=True,
+        cache_type="readahead",
+        cache_options=None,
+        client_kwargs=None,
+        encoded=False,
+        **storage_options,
+    ):
+        """
+
+        Parameters
+        ----------
+        block_size: int
+            Blocks to read bytes; if 0, will default to raw requests file-like
+            objects instead of HTTPFile instances
+        simple_links: bool
+            If True, will consider both HTML <a> tags and anything that looks
+            like a URL; if False, will consider only the former.
+        same_scheme: True
+            When doing ls/glob, if this is True, only consider paths that have
+            http/https matching the input URLs.
+        size_policy: this argument is deprecated
+        client_kwargs: dict
+            Passed to aiohttp.ClientSession, see
+            https://docs.aiohttp.org/en/stable/client_reference.html
+            For example, ``{'auth': aiohttp.BasicAuth('user', 'pass')}``
+        storage_options: key-value
+            Any other parameters passed on to requests
+        cache_type, cache_options: defaults used in open
+        """
+        super().__init__(self, **storage_options)
+        self.block_size = block_size if block_size is not None else DEFAULT_BLOCK_SIZE
+        self.simple_links = simple_links
+        self.same_schema = same_scheme
+        self.cache_type = cache_type
+        self.cache_options = cache_options
+        self.client_kwargs = client_kwargs or {}
+        self.encoded = encoded
+        self.kwargs = storage_options
+
+        try:
+            import js  # noqa: F401
+
+            logger.debug("Starting JS session")
+            self.session = RequestsSessionShim()
+            self.js = True
+        except Exception as e:
+            import requests
+
+            logger.debug("Starting cpython session because of: %s", e)
+            self.session = requests.Session(**(client_kwargs or {}))
+            self.js = False
+
+        request_options = copy(storage_options)
+        self.use_listings_cache = request_options.pop("use_listings_cache", False)
+        request_options.pop("listings_expiry_time", None)
+        request_options.pop("max_paths", None)
+        request_options.pop("skip_instance_cache", None)
+        self.kwargs = request_options
+
+    @property
+    def fsid(self):
+        return "sync-http"
+
+    def encode_url(self, url):
+        if yarl:
+            return yarl.URL(url, encoded=self.encoded)
+        return url
+
+    @classmethod
+    def _strip_protocol(cls, path: str) -> str:
+        """For HTTP, we always want to keep the full URL"""
+        path = path.replace("sync-http://", "http://").replace(
+            "sync-https://", "https://"
+        )
+        return path
+
+    @classmethod
+    def _parent(cls, path):
+        # override, since _strip_protocol is different for URLs
+        par = super()._parent(path)
+        if len(par) > 7:  # "http://..."
+            return par
+        return ""
+
+    def _ls_real(self, url, detail=True, **kwargs):
+        # ignoring URL-encoded arguments
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        logger.debug(url)
+        r = self.session.get(self.encode_url(url), **self.kwargs)
+        self._raise_not_found_for_status(r, url)
+        text = r.text
+        if self.simple_links:
+            links = ex2.findall(text) + [u[2] for u in ex.findall(text)]
+        else:
+            links = [u[2] for u in ex.findall(text)]
+        out = set()
+        parts = urlparse(url)
+        for l in links:
+            if isinstance(l, tuple):
+                l = l[1]
+            if l.startswith("/") and len(l) > 1:
+                # absolute URL on this server
+                l = parts.scheme + "://" + parts.netloc + l
+            if l.startswith("http"):
+                if self.same_schema and l.startswith(url.rstrip("/") + "/"):
+                    out.add(l)
+                elif l.replace("https", "http").startswith(
+                    url.replace("https", "http").rstrip("/") + "/"
+                ):
+                    # allowed to cross http <-> https
+                    out.add(l)
+            else:
+                if l not in ["..", "../"]:
+                    # Ignore FTP-like "parent"
+                    out.add("/".join([url.rstrip("/"), l.lstrip("/")]))
+        if not out and url.endswith("/"):
+            out = self._ls_real(url.rstrip("/"), detail=False)
+        if detail:
+            return [
+                {
+                    "name": u,
+                    "size": None,
+                    "type": "directory" if u.endswith("/") else "file",
+                }
+                for u in out
+            ]
+        else:
+            return sorted(out)
+
+    def ls(self, url, detail=True, **kwargs):
+        if self.use_listings_cache and url in self.dircache:
+            out = self.dircache[url]
+        else:
+            out = self._ls_real(url, detail=detail, **kwargs)
+            self.dircache[url] = out
+        return out
+
+    def _raise_not_found_for_status(self, response, url):
+        """
+        Raises FileNotFoundError for 404s, otherwise uses raise_for_status.
+        """
+        if response.status_code == 404:
+            raise FileNotFoundError(url)
+        response.raise_for_status()
+
+    def cat_file(self, url, start=None, end=None, **kwargs):
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        logger.debug(url)
+
+        if start is not None or end is not None:
+            if start == end:
+                return b""
+            headers = kw.pop("headers", {}).copy()
+
+            headers["Range"] = self._process_limits(url, start, end)
+            kw["headers"] = headers
+        r = self.session.get(self.encode_url(url), **kw)
+        self._raise_not_found_for_status(r, url)
+        return r.content
+
+    def get_file(
+        self, rpath, lpath, chunk_size=5 * 2**20, callback=_DEFAULT_CALLBACK, **kwargs
+    ):
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        logger.debug(rpath)
+        r = self.session.get(self.encode_url(rpath), **kw)
+        try:
+            size = int(
+                r.headers.get("content-length", None)
+                or r.headers.get("Content-Length", None)
+            )
+        except (ValueError, KeyError, TypeError):
+            size = None
+
+        callback.set_size(size)
+        self._raise_not_found_for_status(r, rpath)
+        if not isfilelike(lpath):
+            lpath = open(lpath, "wb")
+        for chunk in r.iter_content(chunk_size, decode_unicode=False):
+            lpath.write(chunk)
+            callback.relative_update(len(chunk))
+
+    def put_file(
+        self,
+        lpath,
+        rpath,
+        chunk_size=5 * 2**20,
+        callback=_DEFAULT_CALLBACK,
+        method="post",
+        **kwargs,
+    ):
+        def gen_chunks():
+            # Support passing arbitrary file-like objects
+            # and use them instead of streams.
+            if isinstance(lpath, io.IOBase):
+                context = nullcontext(lpath)
+                use_seek = False  # might not support seeking
+            else:
+                context = open(lpath, "rb")
+                use_seek = True
+
+            with context as f:
+                if use_seek:
+                    callback.set_size(f.seek(0, 2))
+                    f.seek(0)
+                else:
+                    callback.set_size(getattr(f, "size", None))
+
+                chunk = f.read(chunk_size)
+                while chunk:
+                    yield chunk
+                    callback.relative_update(len(chunk))
+                    chunk = f.read(chunk_size)
+
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+
+        method = method.lower()
+        if method not in ("post", "put"):
+            raise ValueError(
+                f"method has to be either 'post' or 'put', not: {method!r}"
+            )
+
+        meth = getattr(self.session, method)
+        resp = meth(rpath, data=gen_chunks(), **kw)
+        self._raise_not_found_for_status(resp, rpath)
+
+    def _process_limits(self, url, start, end):
+        """Helper for "Range"-based _cat_file"""
+        size = None
+        suff = False
+        if start is not None and start < 0:
+            # if start is negative and end None, end is the "suffix length"
+            if end is None:
+                end = -start
+                start = ""
+                suff = True
+            else:
+                size = size or self.info(url)["size"]
+                start = size + start
+        elif start is None:
+            start = 0
+        if not suff:
+            if end is not None and end < 0:
+                if start is not None:
+                    size = size or self.info(url)["size"]
+                    end = size + end
+            elif end is None:
+                end = ""
+            if isinstance(end, int):
+                end -= 1  # bytes range is inclusive
+        return f"bytes={start}-{end}"
+
+    def exists(self, path, **kwargs):
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        try:
+            logger.debug(path)
+            r = self.session.get(self.encode_url(path), **kw)
+            return r.status_code < 400
+        except Exception:
+            return False
+
+    def isfile(self, path, **kwargs):
+        return self.exists(path, **kwargs)
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        autocommit=None,  # XXX: This differs from the base class.
+        cache_type=None,
+        cache_options=None,
+        size=None,
+        **kwargs,
+    ):
+        """Make a file-like object
+
+        Parameters
+        ----------
+        path: str
+            Full URL with protocol
+        mode: string
+            must be "rb"
+        block_size: int or None
+            Bytes to download in one request; use instance value if None. If
+            zero, will return a streaming Requests file-like instance.
+        kwargs: key-value
+            Any other parameters, passed to requests calls
+        """
+        if mode != "rb":
+            raise NotImplementedError
+        block_size = block_size if block_size is not None else self.block_size
+        kw = self.kwargs.copy()
+        kw.update(kwargs)
+        size = size or self.info(path, **kwargs)["size"]
+        if block_size and size:
+            return HTTPFile(
+                self,
+                path,
+                session=self.session,
+                block_size=block_size,
+                mode=mode,
+                size=size,
+                cache_type=cache_type or self.cache_type,
+                cache_options=cache_options or self.cache_options,
+                **kw,
+            )
+        else:
+            return HTTPStreamFile(
+                self,
+                path,
+                mode=mode,
+                session=self.session,
+                **kw,
+            )
+
+    def ukey(self, url):
+        """Unique identifier; assume HTTP files are static, unchanging"""
+        return tokenize(url, self.kwargs, self.protocol)
+
+    def info(self, url, **kwargs):
+        """Get info of URL
+
+        Tries to access location via HEAD, and then GET methods, but does
+        not fetch the data.
+
+        It is possible that the server does not supply any size information, in
+        which case size will be given as None (and certain operations on the
+        corresponding file will not work).
+        """
+        info = {}
+        for policy in ["head", "get"]:
+            try:
+                info.update(
+                    _file_info(
+                        self.encode_url(url),
+                        size_policy=policy,
+                        session=self.session,
+                        **self.kwargs,
+                        **kwargs,
+                    )
+                )
+                if info.get("size") is not None:
+                    break
+            except Exception as exc:
+                if policy == "get":
+                    # If get failed, then raise a FileNotFoundError
+                    raise FileNotFoundError(url) from exc
+                logger.debug(str(exc))
+
+        return {"name": url, "size": None, **info, "type": "file"}
+
+    def glob(self, path, maxdepth=None, **kwargs):
+        """
+        Find files by glob-matching.
+
+        This implementation is idntical to the one in AbstractFileSystem,
+        but "?" is not considered as a character for globbing, because it is
+        so common in URLs, often identifying the "query" part.
+        """
+        import re
+
+        ends = path.endswith("/")
+        path = self._strip_protocol(path)
+        indstar = path.find("*") if path.find("*") >= 0 else len(path)
+        indbrace = path.find("[") if path.find("[") >= 0 else len(path)
+
+        ind = min(indstar, indbrace)
+
+        detail = kwargs.pop("detail", False)
+
+        if not has_magic(path):
+            root = path
+            depth = 1
+            if ends:
+                path += "/*"
+            elif self.exists(path):
+                if not detail:
+                    return [path]
+                else:
+                    return {path: self.info(path)}
+            else:
+                if not detail:
+                    return []  # glob of non-existent returns empty
+                else:
+                    return {}
+        elif "/" in path[:ind]:
+            ind2 = path[:ind].rindex("/")
+            root = path[: ind2 + 1]
+            depth = None if "**" in path else path[ind2 + 1 :].count("/") + 1
+        else:
+            root = ""
+            depth = None if "**" in path else path[ind + 1 :].count("/") + 1
+
+        allpaths = self.find(
+            root, maxdepth=maxdepth or depth, withdirs=True, detail=True, **kwargs
+        )
+        # Escape characters special to python regex, leaving our supported
+        # special characters in place.
+        # See https://www.gnu.org/software/bash/manual/html_node/Pattern-Matching.html
+        # for shell globbing details.
+        pattern = (
+            "^"
+            + (
+                path.replace("\\", r"\\")
+                .replace(".", r"\.")
+                .replace("+", r"\+")
+                .replace("//", "/")
+                .replace("(", r"\(")
+                .replace(")", r"\)")
+                .replace("|", r"\|")
+                .replace("^", r"\^")
+                .replace("$", r"\$")
+                .replace("{", r"\{")
+                .replace("}", r"\}")
+                .rstrip("/")
+            )
+            + "$"
+        )
+        pattern = re.sub("[*]{2}", "=PLACEHOLDER=", pattern)
+        pattern = re.sub("[*]", "[^/]*", pattern)
+        pattern = re.compile(pattern.replace("=PLACEHOLDER=", ".*"))
+        out = {
+            p: allpaths[p]
+            for p in sorted(allpaths)
+            if pattern.match(p.replace("//", "/").rstrip("/"))
+        }
+        if detail:
+            return out
+        else:
+            return list(out)
+
+    def isdir(self, path):
+        # override, since all URLs are (also) files
+        try:
+            return bool(self.ls(path))
+        except (FileNotFoundError, ValueError):
+            return False
+
+
+class HTTPFile(AbstractBufferedFile):
+    """
+    A file-like object pointing to a remove HTTP(S) resource
+
+    Supports only reading, with read-ahead of a predermined block-size.
+
+    In the case that the server does not supply the filesize, only reading of
+    the complete file in one go is supported.
+
+    Parameters
+    ----------
+    url: str
+        Full URL of the remote resource, including the protocol
+    session: requests.Session or None
+        All calls will be made within this session, to avoid restarting
+        connections where the server allows this
+    block_size: int or None
+        The amount of read-ahead to do, in bytes. Default is 5MB, or the value
+        configured for the FileSystem creating this file
+    size: None or int
+        If given, this is the size of the file in bytes, and we don't attempt
+        to call the server to find the value.
+    kwargs: all other key-values are passed to requests calls.
+    """
+
+    def __init__(
+        self,
+        fs,
+        url,
+        session=None,
+        block_size=None,
+        mode="rb",
+        cache_type="bytes",
+        cache_options=None,
+        size=None,
+        **kwargs,
+    ):
+        if mode != "rb":
+            raise NotImplementedError("File mode not supported")
+        self.url = url
+        self.session = session
+        self.details = {"name": url, "size": size, "type": "file"}
+        super().__init__(
+            fs=fs,
+            path=url,
+            mode=mode,
+            block_size=block_size,
+            cache_type=cache_type,
+            cache_options=cache_options,
+            **kwargs,
+        )
+
+    def read(self, length=-1):
+        """Read bytes from file
+
+        Parameters
+        ----------
+        length: int
+            Read up to this many bytes. If negative, read all content to end of
+            file. If the server has not supplied the filesize, attempting to
+            read only part of the data will raise a ValueError.
+        """
+        if (
+            (length < 0 and self.loc == 0)  # explicit read all
+            # but not when the size is known and fits into a block anyways
+            and not (self.size is not None and self.size <= self.blocksize)
+        ):
+            self._fetch_all()
+        if self.size is None:
+            if length < 0:
+                self._fetch_all()
+        else:
+            length = min(self.size - self.loc, length)
+        return super().read(length)
+
+    def _fetch_all(self):
+        """Read whole file in one shot, without caching
+
+        This is only called when position is still at zero,
+        and read() is called without a byte-count.
+        """
+        logger.debug(f"Fetch all for {self}")
+        if not isinstance(self.cache, AllBytes):
+            r = self.session.get(self.fs.encode_url(self.url), **self.kwargs)
+            r.raise_for_status()
+            out = r.content
+            self.cache = AllBytes(size=len(out), fetcher=None, blocksize=None, data=out)
+            self.size = len(out)
+
+    def _parse_content_range(self, headers):
+        """Parse the Content-Range header"""
+        s = headers.get("Content-Range", "")
+        m = re.match(r"bytes (\d+-\d+|\*)/(\d+|\*)", s)
+        if not m:
+            return None, None, None
+
+        if m[1] == "*":
+            start = end = None
+        else:
+            start, end = [int(x) for x in m[1].split("-")]
+        total = None if m[2] == "*" else int(m[2])
+        return start, end, total
+
+    def _fetch_range(self, start, end):
+        """Download a block of data
+
+        The expectation is that the server returns only the requested bytes,
+        with HTTP code 206. If this is not the case, we first check the headers,
+        and then stream the output - if the data size is bigger than we
+        requested, an exception is raised.
+        """
+        logger.debug(f"Fetch range for {self}: {start}-{end}")
+        kwargs = self.kwargs.copy()
+        headers = kwargs.pop("headers", {}).copy()
+        headers["Range"] = f"bytes={start}-{end - 1}"
+        logger.debug("%s : %s", self.url, headers["Range"])
+        r = self.session.get(self.fs.encode_url(self.url), headers=headers, **kwargs)
+        if r.status_code == 416:
+            # range request outside file
+            return b""
+        r.raise_for_status()
+
+        # If the server has handled the range request, it should reply
+        # with status 206 (partial content). But we'll guess that a suitable
+        # Content-Range header or a Content-Length no more than the
+        # requested range also mean we have got the desired range.
+        cl = r.headers.get("Content-Length", r.headers.get("content-length", end + 1))
+        response_is_range = (
+            r.status_code == 206
+            or self._parse_content_range(r.headers)[0] == start
+            or int(cl) <= end - start
+        )
+
+        if response_is_range:
+            # partial content, as expected
+            out = r.content
+        elif start > 0:
+            raise ValueError(
+                "The HTTP server doesn't appear to support range requests. "
+                "Only reading this file from the beginning is supported. "
+                "Open with block_size=0 for a streaming file interface."
+            )
+        else:
+            # Response is not a range, but we want the start of the file,
+            # so we can read the required amount anyway.
+            cl = 0
+            out = []
+            for chunk in r.iter_content(2**20, False):
+                out.append(chunk)
+                cl += len(chunk)
+            out = b"".join(out)[: end - start]
+        return out
+
+
+magic_check = re.compile("([*[])")
+
+
+def has_magic(s):
+    match = magic_check.search(s)
+    return match is not None
+
+
+class HTTPStreamFile(AbstractBufferedFile):
+    def __init__(self, fs, url, mode="rb", session=None, **kwargs):
+        self.url = url
+        self.session = session
+        if mode != "rb":
+            raise ValueError
+        self.details = {"name": url, "size": None}
+        super().__init__(fs=fs, path=url, mode=mode, cache_type="readahead", **kwargs)
+
+        r = self.session.get(self.fs.encode_url(url), stream=True, **kwargs)
+        self.fs._raise_not_found_for_status(r, url)
+        self.it = r.iter_content(1024, False)
+        self.leftover = b""
+
+        self.r = r
+
+    def seek(self, *args, **kwargs):
+        raise ValueError("Cannot seek streaming HTTP file")
+
+    def read(self, num=-1):
+        bufs = [self.leftover]
+        leng = len(self.leftover)
+        while leng < num or num < 0:
+            try:
+                out = self.it.__next__()
+            except StopIteration:
+                break
+            if out:
+                bufs.append(out)
+            else:
+                break
+            leng += len(out)
+        out = b"".join(bufs)
+        if num >= 0:
+            self.leftover = out[num:]
+            out = out[:num]
+        else:
+            self.leftover = b""
+        self.loc += len(out)
+        return out
+
+    def close(self):
+        self.r.close()
+        self.closed = True
+
+
+def get_range(session, url, start, end, **kwargs):
+    # explicit get a range when we know it must be safe
+    kwargs = kwargs.copy()
+    headers = kwargs.pop("headers", {}).copy()
+    headers["Range"] = f"bytes={start}-{end - 1}"
+    r = session.get(url, headers=headers, **kwargs)
+    r.raise_for_status()
+    return r.content
+
+
+def _file_info(url, session, size_policy="head", **kwargs):
+    """Call HEAD on the server to get details about the file (size/checksum etc.)
+
+    Default operation is to explicitly allow redirects and use encoding
+    'identity' (no compression) to get the true size of the target.
+    """
+    logger.debug("Retrieve file size for %s", url)
+    kwargs = kwargs.copy()
+    ar = kwargs.pop("allow_redirects", True)
+    head = kwargs.get("headers", {}).copy()
+    # TODO: not allowed in JS
+    # head["Accept-Encoding"] = "identity"
+    kwargs["headers"] = head
+
+    info = {}
+    if size_policy == "head":
+        r = session.head(url, allow_redirects=ar, **kwargs)
+    elif size_policy == "get":
+        r = session.get(url, allow_redirects=ar, **kwargs)
+    else:
+        raise TypeError(f'size_policy must be "head" or "get", got {size_policy}')
+    r.raise_for_status()
+
+    # TODO:
+    #  recognise lack of 'Accept-Ranges',
+    #                 or 'Accept-Ranges': 'none' (not 'bytes')
+    #  to mean streaming only, no random access => return None
+    if "Content-Length" in r.headers:
+        info["size"] = int(r.headers["Content-Length"])
+    elif "Content-Range" in r.headers:
+        info["size"] = int(r.headers["Content-Range"].split("/")[1])
+    elif "content-length" in r.headers:
+        info["size"] = int(r.headers["content-length"])
+    elif "content-range" in r.headers:
+        info["size"] = int(r.headers["content-range"].split("/")[1])
+
+    for checksum_field in ["ETag", "Content-MD5", "Digest"]:
+        if r.headers.get(checksum_field):
+            info[checksum_field] = r.headers[checksum_field]
+
+    return info
+
+
+# importing this is enough to register it
+def register():
+    register_implementation("http", HTTPFileSystem, clobber=True)
+    register_implementation("https", HTTPFileSystem, clobber=True)
+    register_implementation("sync-http", HTTPFileSystem, clobber=True)
+    register_implementation("sync-https", HTTPFileSystem, clobber=True)
+
+
+register()
+
+
+def unregister():
+    from fsspec.implementations.http import HTTPFileSystem
+
+    register_implementation("http", HTTPFileSystem, clobber=True)
+    register_implementation("https", HTTPFileSystem, clobber=True)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/jupyter.py

@@ -0,0 +1,129 @@
+import base64
+import io
+import re
+
+import requests
+
+import fsspec
+
+
+class JupyterFileSystem(fsspec.AbstractFileSystem):
+    """View of the files as seen by a Jupyter server (notebook or lab)"""
+
+    protocol = ("jupyter", "jlab")
+
+    def __init__(self, url, tok=None, **kwargs):
+        """
+
+        Parameters
+        ----------
+        url : str
+            Base URL of the server, like "http://127.0.0.1:8888". May include
+            token in the string, which is given by the process when starting up
+        tok : str
+            If the token is obtained separately, can be given here
+        kwargs
+        """
+        if "?" in url:
+            if tok is None:
+                try:
+                    tok = re.findall("token=([a-z0-9]+)", url)[0]
+                except IndexError as e:
+                    raise ValueError("Could not determine token") from e
+            url = url.split("?", 1)[0]
+        self.url = url.rstrip("/") + "/api/contents"
+        self.session = requests.Session()
+        if tok:
+            self.session.headers["Authorization"] = f"token {tok}"
+
+        super().__init__(**kwargs)
+
+    def ls(self, path, detail=True, **kwargs):
+        path = self._strip_protocol(path)
+        r = self.session.get(f"{self.url}/{path}")
+        if r.status_code == 404:
+            raise FileNotFoundError(path)
+        r.raise_for_status()
+        out = r.json()
+
+        if out["type"] == "directory":
+            out = out["content"]
+        else:
+            out = [out]
+        for o in out:
+            o["name"] = o.pop("path")
+            o.pop("content")
+            if o["type"] == "notebook":
+                o["type"] = "file"
+        if detail:
+            return out
+        return [o["name"] for o in out]
+
+    def cat_file(self, path, start=None, end=None, **kwargs):
+        path = self._strip_protocol(path)
+        r = self.session.get(f"{self.url}/{path}")
+        if r.status_code == 404:
+            raise FileNotFoundError(path)
+        r.raise_for_status()
+        out = r.json()
+        if out["format"] == "text":
+            # data should be binary
+            b = out["content"].encode()
+        else:
+            b = base64.b64decode(out["content"])
+        return b[start:end]
+
+    def pipe_file(self, path, value, **_):
+        path = self._strip_protocol(path)
+        json = {
+            "name": path.rsplit("/", 1)[-1],
+            "path": path,
+            "size": len(value),
+            "content": base64.b64encode(value).decode(),
+            "format": "base64",
+            "type": "file",
+        }
+        self.session.put(f"{self.url}/{path}", json=json)
+
+    def mkdir(self, path, create_parents=True, **kwargs):
+        path = self._strip_protocol(path)
+        if create_parents and "/" in path:
+            self.mkdir(path.rsplit("/", 1)[0], True)
+        json = {
+            "name": path.rsplit("/", 1)[-1],
+            "path": path,
+            "size": None,
+            "content": None,
+            "type": "directory",
+        }
+        self.session.put(f"{self.url}/{path}", json=json)
+
+    def mv(self, path1, path2, recursive=False, maxdepth=None, **kwargs):
+        if path1 == path2:
+            return
+        self.session.patch(f"{self.url}/{path1}", json={"path": path2})
+
+    def _rm(self, path):
+        path = self._strip_protocol(path)
+        self.session.delete(f"{self.url}/{path}")
+
+    def _open(self, path, mode="rb", **kwargs):
+        path = self._strip_protocol(path)
+        if mode == "rb":
+            data = self.cat_file(path)
+            return io.BytesIO(data)
+        else:
+            return SimpleFileWriter(self, path, mode="wb")
+
+
+class SimpleFileWriter(fsspec.spec.AbstractBufferedFile):
+    def _upload_chunk(self, final=False):
+        """Never uploads a chunk until file is done
+
+        Not suitable for large files
+        """
+        if final is False:
+            return False
+        self.buffer.seek(0)
+        data = self.buffer.read()
+        self.fs.pipe_file(self.path, data)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/libarchive.py

@@ -0,0 +1,213 @@
+from contextlib import contextmanager
+from ctypes import (
+    CFUNCTYPE,
+    POINTER,
+    c_int,
+    c_longlong,
+    c_void_p,
+    cast,
+    create_string_buffer,
+)
+
+import libarchive
+import libarchive.ffi as ffi
+
+from fsspec import open_files
+from fsspec.archive import AbstractArchiveFileSystem
+from fsspec.implementations.memory import MemoryFile
+from fsspec.utils import DEFAULT_BLOCK_SIZE
+
+# Libarchive requires seekable files or memory only for certain archive
+# types. However, since we read the directory first to cache the contents
+# and also allow random access to any file, the file-like object needs
+# to be seekable no matter what.
+
+# Seek call-backs (not provided in the libarchive python wrapper)
+SEEK_CALLBACK = CFUNCTYPE(c_longlong, c_int, c_void_p, c_longlong, c_int)
+read_set_seek_callback = ffi.ffi(
+    "read_set_seek_callback", [ffi.c_archive_p, SEEK_CALLBACK], c_int, ffi.check_int
+)
+new_api = hasattr(ffi, "NO_OPEN_CB")
+
+
+@contextmanager
+def custom_reader(file, format_name="all", filter_name="all", block_size=ffi.page_size):
+    """Read an archive from a seekable file-like object.
+
+    The `file` object must support the standard `readinto` and 'seek' methods.
+    """
+    buf = create_string_buffer(block_size)
+    buf_p = cast(buf, c_void_p)
+
+    def read_func(archive_p, context, ptrptr):
+        # readinto the buffer, returns number of bytes read
+        length = file.readinto(buf)
+        # write the address of the buffer into the pointer
+        ptrptr = cast(ptrptr, POINTER(c_void_p))
+        ptrptr[0] = buf_p
+        # tell libarchive how much data was written into the buffer
+        return length
+
+    def seek_func(archive_p, context, offset, whence):
+        file.seek(offset, whence)
+        # tell libarchvie the current position
+        return file.tell()
+
+    read_cb = ffi.READ_CALLBACK(read_func)
+    seek_cb = SEEK_CALLBACK(seek_func)
+
+    if new_api:
+        open_cb = ffi.NO_OPEN_CB
+        close_cb = ffi.NO_CLOSE_CB
+    else:
+        open_cb = libarchive.read.OPEN_CALLBACK(ffi.VOID_CB)
+        close_cb = libarchive.read.CLOSE_CALLBACK(ffi.VOID_CB)
+
+    with libarchive.read.new_archive_read(format_name, filter_name) as archive_p:
+        read_set_seek_callback(archive_p, seek_cb)
+        ffi.read_open(archive_p, None, open_cb, read_cb, close_cb)
+        yield libarchive.read.ArchiveRead(archive_p)
+
+
+class LibArchiveFileSystem(AbstractArchiveFileSystem):
+    """Compressed archives as a file-system (read-only)
+
+    Supports the following formats:
+    tar, pax , cpio, ISO9660, zip, mtree, shar, ar, raw, xar, lha/lzh, rar
+    Microsoft CAB, 7-Zip, WARC
+
+    See the libarchive documentation for further restrictions.
+    https://www.libarchive.org/
+
+    Keeps file object open while instance lives. It only works in seekable
+    file-like objects. In case the filesystem does not support this kind of
+    file object, it is recommended to cache locally.
+
+    This class is pickleable, but not necessarily thread-safe (depends on the
+    platform). See libarchive documentation for details.
+    """
+
+    root_marker = ""
+    protocol = "libarchive"
+    cachable = False
+
+    def __init__(
+        self,
+        fo="",
+        mode="r",
+        target_protocol=None,
+        target_options=None,
+        block_size=DEFAULT_BLOCK_SIZE,
+        **kwargs,
+    ):
+        """
+        Parameters
+        ----------
+        fo: str or file-like
+            Contains ZIP, and must exist. If a str, will fetch file using
+            :meth:`~fsspec.open_files`, which must return one file exactly.
+        mode: str
+            Currently, only 'r' accepted
+        target_protocol: str (optional)
+            If ``fo`` is a string, this value can be used to override the
+            FS protocol inferred from a URL
+        target_options: dict (optional)
+            Kwargs passed when instantiating the target FS, if ``fo`` is
+            a string.
+        """
+        super().__init__(self, **kwargs)
+        if mode != "r":
+            raise ValueError("Only read from archive files accepted")
+        if isinstance(fo, str):
+            files = open_files(fo, protocol=target_protocol, **(target_options or {}))
+            if len(files) != 1:
+                raise ValueError(
+                    f'Path "{fo}" did not resolve to exactly one file: "{files}"'
+                )
+            fo = files[0]
+        self.of = fo
+        self.fo = fo.__enter__()  # the whole instance is a context
+        self.block_size = block_size
+        self.dir_cache = None
+
+    @contextmanager
+    def _open_archive(self):
+        self.fo.seek(0)
+        with custom_reader(self.fo, block_size=self.block_size) as arc:
+            yield arc
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        # file paths are always relative to the archive root
+        return super()._strip_protocol(path).lstrip("/")
+
+    def _get_dirs(self):
+        fields = {
+            "name": "pathname",
+            "size": "size",
+            "created": "ctime",
+            "mode": "mode",
+            "uid": "uid",
+            "gid": "gid",
+            "mtime": "mtime",
+        }
+
+        if self.dir_cache is not None:
+            return
+
+        self.dir_cache = {}
+        list_names = []
+        with self._open_archive() as arc:
+            for entry in arc:
+                if not entry.isdir and not entry.isfile:
+                    # Skip symbolic links, fifo entries, etc.
+                    continue
+                self.dir_cache.update(
+                    {
+                        dirname: {"name": dirname, "size": 0, "type": "directory"}
+                        for dirname in self._all_dirnames(set(entry.name))
+                    }
+                )
+                f = {key: getattr(entry, fields[key]) for key in fields}
+                f["type"] = "directory" if entry.isdir else "file"
+                list_names.append(entry.name)
+
+                self.dir_cache[f["name"]] = f
+        # libarchive does not seem to return an entry for the directories (at least
+        # not in all formats), so get the directories names from the files names
+        self.dir_cache.update(
+            {
+                dirname: {"name": dirname, "size": 0, "type": "directory"}
+                for dirname in self._all_dirnames(list_names)
+            }
+        )
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        autocommit=True,
+        cache_options=None,
+        **kwargs,
+    ):
+        path = self._strip_protocol(path)
+        if mode != "rb":
+            raise NotImplementedError
+
+        data = bytes()
+        with self._open_archive() as arc:
+            for entry in arc:
+                if entry.pathname != path:
+                    continue
+
+                if entry.size == 0:
+                    # empty file, so there are no blocks
+                    break
+
+                for block in entry.get_blocks(entry.size):
+                    data = block
+                    break
+                else:
+                    raise ValueError
+        return MemoryFile(fs=self, path=path, data=data)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/local.py

@@ -0,0 +1,514 @@
+import datetime
+import io
+import logging
+import os
+import os.path as osp
+import shutil
+import stat
+import tempfile
+from functools import lru_cache
+
+from fsspec import AbstractFileSystem
+from fsspec.compression import compr
+from fsspec.core import get_compression
+from fsspec.utils import isfilelike, stringify_path
+
+logger = logging.getLogger("fsspec.local")
+
+
+class LocalFileSystem(AbstractFileSystem):
+    """Interface to files on local storage
+
+    Parameters
+    ----------
+    auto_mkdir: bool
+        Whether, when opening a file, the directory containing it should
+        be created (if it doesn't already exist). This is assumed by pyarrow
+        code.
+    """
+
+    root_marker = "/"
+    protocol = "file", "local"
+    local_file = True
+
+    def __init__(self, auto_mkdir=False, **kwargs):
+        super().__init__(**kwargs)
+        self.auto_mkdir = auto_mkdir
+
+    @property
+    def fsid(self):
+        return "local"
+
+    def mkdir(self, path, create_parents=True, **kwargs):
+        path = self._strip_protocol(path)
+        if self.exists(path):
+            raise FileExistsError(path)
+        if create_parents:
+            self.makedirs(path, exist_ok=True)
+        else:
+            os.mkdir(path, **kwargs)
+
+    def makedirs(self, path, exist_ok=False):
+        path = self._strip_protocol(path)
+        os.makedirs(path, exist_ok=exist_ok)
+
+    def rmdir(self, path):
+        path = self._strip_protocol(path)
+        os.rmdir(path)
+
+    def ls(self, path, detail=False, **kwargs):
+        path = self._strip_protocol(path)
+        path_info = self.info(path)
+        infos = []
+        if path_info["type"] == "directory":
+            with os.scandir(path) as it:
+                for f in it:
+                    try:
+                        # Only get the info if requested since it is a bit expensive (the stat call inside)
+                        # The strip_protocol is also used in info() and calls make_path_posix to always return posix paths
+                        info = self.info(f) if detail else self._strip_protocol(f.path)
+                        infos.append(info)
+                    except FileNotFoundError:
+                        pass
+        else:
+            infos = [path_info] if detail else [path_info["name"]]
+
+        return infos
+
+    def info(self, path, **kwargs):
+        if isinstance(path, os.DirEntry):
+            # scandir DirEntry
+            out = path.stat(follow_symlinks=False)
+            link = path.is_symlink()
+            if path.is_dir(follow_symlinks=False):
+                t = "directory"
+            elif path.is_file(follow_symlinks=False):
+                t = "file"
+            else:
+                t = "other"
+
+            size = out.st_size
+            if link:
+                try:
+                    out2 = path.stat(follow_symlinks=True)
+                    size = out2.st_size
+                except OSError:
+                    size = 0
+            path = self._strip_protocol(path.path)
+        else:
+            # str or path-like
+            path = self._strip_protocol(path)
+            out = os.stat(path, follow_symlinks=False)
+            link = stat.S_ISLNK(out.st_mode)
+            if link:
+                out = os.stat(path, follow_symlinks=True)
+            size = out.st_size
+            if stat.S_ISDIR(out.st_mode):
+                t = "directory"
+            elif stat.S_ISREG(out.st_mode):
+                t = "file"
+            else:
+                t = "other"
+
+        # Check for the 'st_birthtime' attribute, which is not always present; fallback to st_ctime
+        created_time = getattr(out, "st_birthtime", out.st_ctime)
+
+        result = {
+            "name": path,
+            "size": size,
+            "type": t,
+            "created": created_time,
+            "islink": link,
+        }
+        for field in ["mode", "uid", "gid", "mtime", "ino", "nlink"]:
+            result[field] = getattr(out, f"st_{field}")
+        if link:
+            result["destination"] = os.readlink(path)
+        return result
+
+    def lexists(self, path, **kwargs):
+        return osp.lexists(path)
+
+    def cp_file(self, path1, path2, **kwargs):
+        path1 = self._strip_protocol(path1)
+        path2 = self._strip_protocol(path2)
+        if self.auto_mkdir:
+            self.makedirs(self._parent(path2), exist_ok=True)
+        if self.isfile(path1):
+            shutil.copyfile(path1, path2)
+        elif self.isdir(path1):
+            self.mkdirs(path2, exist_ok=True)
+        else:
+            raise FileNotFoundError(path1)
+
+    def isfile(self, path):
+        path = self._strip_protocol(path)
+        return os.path.isfile(path)
+
+    def isdir(self, path):
+        path = self._strip_protocol(path)
+        return os.path.isdir(path)
+
+    def get_file(self, path1, path2, callback=None, **kwargs):
+        if isfilelike(path2):
+            with open(path1, "rb") as f:
+                shutil.copyfileobj(f, path2)
+        else:
+            return self.cp_file(path1, path2, **kwargs)
+
+    def put_file(self, path1, path2, callback=None, **kwargs):
+        return self.cp_file(path1, path2, **kwargs)
+
+    def mv(self, path1, path2, recursive: bool = True, **kwargs):
+        """Move files/directories
+        For the specific case of local, all ops on directories are recursive and
+        the recursive= kwarg is ignored.
+        """
+        path1 = self._strip_protocol(path1)
+        path2 = self._strip_protocol(path2)
+        shutil.move(path1, path2)
+
+    def link(self, src, dst, **kwargs):
+        src = self._strip_protocol(src)
+        dst = self._strip_protocol(dst)
+        os.link(src, dst, **kwargs)
+
+    def symlink(self, src, dst, **kwargs):
+        src = self._strip_protocol(src)
+        dst = self._strip_protocol(dst)
+        os.symlink(src, dst, **kwargs)
+
+    def islink(self, path) -> bool:
+        return os.path.islink(self._strip_protocol(path))
+
+    def rm_file(self, path):
+        os.remove(self._strip_protocol(path))
+
+    def rm(self, path, recursive=False, maxdepth=None):
+        if not isinstance(path, list):
+            path = [path]
+
+        for p in path:
+            p = self._strip_protocol(p)
+            if self.isdir(p):
+                if not recursive:
+                    raise ValueError("Cannot delete directory, set recursive=True")
+                if osp.abspath(p) == os.getcwd():
+                    raise ValueError("Cannot delete current working directory")
+                shutil.rmtree(p)
+            else:
+                os.remove(p)
+
+    def unstrip_protocol(self, name):
+        name = self._strip_protocol(name)  # normalise for local/win/...
+        return f"file://{name}"
+
+    def _open(self, path, mode="rb", block_size=None, **kwargs):
+        path = self._strip_protocol(path)
+        if self.auto_mkdir and "w" in mode:
+            self.makedirs(self._parent(path), exist_ok=True)
+        return LocalFileOpener(path, mode, fs=self, **kwargs)
+
+    def touch(self, path, truncate=True, **kwargs):
+        path = self._strip_protocol(path)
+        if self.auto_mkdir:
+            self.makedirs(self._parent(path), exist_ok=True)
+        if self.exists(path):
+            os.utime(path, None)
+        else:
+            open(path, "a").close()
+        if truncate:
+            os.truncate(path, 0)
+
+    def created(self, path):
+        info = self.info(path=path)
+        return datetime.datetime.fromtimestamp(
+            info["created"], tz=datetime.timezone.utc
+        )
+
+    def modified(self, path):
+        info = self.info(path=path)
+        return datetime.datetime.fromtimestamp(info["mtime"], tz=datetime.timezone.utc)
+
+    @classmethod
+    def _parent(cls, path):
+        path = cls._strip_protocol(path)
+        if os.sep == "/":
+            # posix native
+            return path.rsplit("/", 1)[0] or "/"
+        else:
+            # NT
+            path_ = path.rsplit("/", 1)[0]
+            if len(path_) <= 3:
+                if path_[1:2] == ":":
+                    # nt root (something like c:/)
+                    return path_[0] + ":/"
+            # More cases may be required here
+            return path_
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        path = stringify_path(path)
+        if path.startswith("file://"):
+            path = path[7:]
+        elif path.startswith("file:"):
+            path = path[5:]
+        elif path.startswith("local://"):
+            path = path[8:]
+        elif path.startswith("local:"):
+            path = path[6:]
+
+        path = make_path_posix(path)
+        if os.sep != "/":
+            # This code-path is a stripped down version of
+            # > drive, path = ntpath.splitdrive(path)
+            if path[1:2] == ":":
+                # Absolute drive-letter path, e.g. X:\Windows
+                # Relative path with drive, e.g. X:Windows
+                drive, path = path[:2], path[2:]
+            elif path[:2] == "//":
+                # UNC drives, e.g. \\server\share or \\?\UNC\server\share
+                # Device drives, e.g. \\.\device or \\?\device
+                if (index1 := path.find("/", 2)) == -1 or (
+                    index2 := path.find("/", index1 + 1)
+                ) == -1:
+                    drive, path = path, ""
+                else:
+                    drive, path = path[:index2], path[index2:]
+            else:
+                # Relative path, e.g. Windows
+                drive = ""
+
+            path = path.rstrip("/") or cls.root_marker
+            return drive + path
+
+        else:
+            return path.rstrip("/") or cls.root_marker
+
+    def _isfilestore(self):
+        # Inheriting from DaskFileSystem makes this False (S3, etc. were)
+        # the original motivation. But we are a posix-like file system.
+        # See https://github.com/dask/dask/issues/5526
+        return True
+
+    def chmod(self, path, mode):
+        path = stringify_path(path)
+        return os.chmod(path, mode)
+
+
+def make_path_posix(path):
+    """Make path generic and absolute for current OS"""
+    if not isinstance(path, str):
+        if isinstance(path, (list, set, tuple)):
+            return type(path)(make_path_posix(p) for p in path)
+        else:
+            path = stringify_path(path)
+            if not isinstance(path, str):
+                raise TypeError(f"could not convert {path!r} to string")
+    if os.sep == "/":
+        # Native posix
+        if path.startswith("/"):
+            # most common fast case for posix
+            return path
+        elif path.startswith("~"):
+            return osp.expanduser(path)
+        elif path.startswith("./"):
+            path = path[2:]
+        elif path == ".":
+            path = ""
+        return f"{os.getcwd()}/{path}"
+    else:
+        # NT handling
+        if path[0:1] == "/" and path[2:3] == ":":
+            # path is like "/c:/local/path"
+            path = path[1:]
+        if path[1:2] == ":":
+            # windows full path like "C:\\local\\path"
+            if len(path) <= 3:
+                # nt root (something like c:/)
+                return path[0] + ":/"
+            path = path.replace("\\", "/")
+            return path
+        elif path[0:1] == "~":
+            return make_path_posix(osp.expanduser(path))
+        elif path.startswith(("\\\\", "//")):
+            # windows UNC/DFS-style paths
+            return "//" + path[2:].replace("\\", "/")
+        elif path.startswith(("\\", "/")):
+            # windows relative path with root
+            path = path.replace("\\", "/")
+            return f"{osp.splitdrive(os.getcwd())[0]}{path}"
+        else:
+            path = path.replace("\\", "/")
+            if path.startswith("./"):
+                path = path[2:]
+            elif path == ".":
+                path = ""
+            return f"{make_path_posix(os.getcwd())}/{path}"
+
+
+def trailing_sep(path):
+    """Return True if the path ends with a path separator.
+
+    A forward slash is always considered a path separator, even on Operating
+    Systems that normally use a backslash.
+    """
+    # TODO: if all incoming paths were posix-compliant then separator would
+    # always be a forward slash, simplifying this function.
+    # See https://github.com/fsspec/filesystem_spec/pull/1250
+    return path.endswith(os.sep) or (os.altsep is not None and path.endswith(os.altsep))
+
+
+@lru_cache(maxsize=1)
+def get_umask(mask: int = 0o666) -> int:
+    """Get the current umask.
+
+    Follows https://stackoverflow.com/a/44130549 to get the umask.
+    Temporarily sets the umask to the given value, and then resets it to the
+    original value.
+    """
+    value = os.umask(mask)
+    os.umask(value)
+    return value
+
+
+class LocalFileOpener(io.IOBase):
+    def __init__(
+        self, path, mode, autocommit=True, fs=None, compression=None, **kwargs
+    ):
+        logger.debug("open file: %s", path)
+        self.path = path
+        self.mode = mode
+        self.fs = fs
+        self.f = None
+        self.autocommit = autocommit
+        self.compression = get_compression(path, compression)
+        self.blocksize = io.DEFAULT_BUFFER_SIZE
+        self._open()
+
+    def _open(self):
+        if self.f is None or self.f.closed:
+            if self.autocommit or "w" not in self.mode:
+                self.f = open(self.path, mode=self.mode)
+                if self.compression:
+                    compress = compr[self.compression]
+                    self.f = compress(self.f, mode=self.mode)
+            else:
+                # TODO: check if path is writable?
+                i, name = tempfile.mkstemp()
+                os.close(i)  # we want normal open and normal buffered file
+                self.temp = name
+                self.f = open(name, mode=self.mode)
+            if "w" not in self.mode:
+                self.size = self.f.seek(0, 2)
+                self.f.seek(0)
+                self.f.size = self.size
+
+    def _fetch_range(self, start, end):
+        # probably only used by cached FS
+        if "r" not in self.mode:
+            raise ValueError
+        self._open()
+        self.f.seek(start)
+        return self.f.read(end - start)
+
+    def __setstate__(self, state):
+        self.f = None
+        loc = state.pop("loc", None)
+        self.__dict__.update(state)
+        if "r" in state["mode"]:
+            self.f = None
+            self._open()
+            self.f.seek(loc)
+
+    def __getstate__(self):
+        d = self.__dict__.copy()
+        d.pop("f")
+        if "r" in self.mode:
+            d["loc"] = self.f.tell()
+        else:
+            if not self.f.closed:
+                raise ValueError("Cannot serialise open write-mode local file")
+        return d
+
+    def commit(self):
+        if self.autocommit:
+            raise RuntimeError("Can only commit if not already set to autocommit")
+        try:
+            shutil.move(self.temp, self.path)
+        except PermissionError as e:
+            # shutil.move raises PermissionError if os.rename
+            # and the default copy2 fallback with shutil.copystats fail.
+            # The file should be there nonetheless, but without copied permissions.
+            # If it doesn't exist, there was no permission to create the file.
+            if not os.path.exists(self.path):
+                raise e
+        else:
+            # If PermissionError is not raised, permissions can be set.
+            try:
+                mask = 0o666
+                os.chmod(self.path, mask & ~get_umask(mask))
+            except RuntimeError:
+                pass
+
+    def discard(self):
+        if self.autocommit:
+            raise RuntimeError("Cannot discard if set to autocommit")
+        os.remove(self.temp)
+
+    def readable(self) -> bool:
+        return True
+
+    def writable(self) -> bool:
+        return "r" not in self.mode
+
+    def read(self, *args, **kwargs):
+        return self.f.read(*args, **kwargs)
+
+    def write(self, *args, **kwargs):
+        return self.f.write(*args, **kwargs)
+
+    def tell(self, *args, **kwargs):
+        return self.f.tell(*args, **kwargs)
+
+    def seek(self, *args, **kwargs):
+        return self.f.seek(*args, **kwargs)
+
+    def seekable(self, *args, **kwargs):
+        return self.f.seekable(*args, **kwargs)
+
+    def readline(self, *args, **kwargs):
+        return self.f.readline(*args, **kwargs)
+
+    def readlines(self, *args, **kwargs):
+        return self.f.readlines(*args, **kwargs)
+
+    def close(self):
+        return self.f.close()
+
+    def truncate(self, size=None) -> int:
+        return self.f.truncate(size)
+
+    @property
+    def closed(self):
+        return self.f.closed
+
+    def fileno(self):
+        return self.raw.fileno()
+
+    def flush(self) -> None:
+        self.f.flush()
+
+    def __iter__(self):
+        return self.f.__iter__()
+
+    def __getattr__(self, item):
+        return getattr(self.f, item)
+
+    def __enter__(self):
+        self._incontext = True
+        return self
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        self._incontext = False
+        self.f.__exit__(exc_type, exc_value, traceback)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/memory.py

@@ -0,0 +1,311 @@
+from __future__ import annotations
+
+import logging
+from datetime import datetime, timezone
+from errno import ENOTEMPTY
+from io import BytesIO
+from pathlib import PurePath, PureWindowsPath
+from typing import Any, ClassVar
+
+from fsspec import AbstractFileSystem
+from fsspec.implementations.local import LocalFileSystem
+from fsspec.utils import stringify_path
+
+logger = logging.getLogger("fsspec.memoryfs")
+
+
+class MemoryFileSystem(AbstractFileSystem):
+    """A filesystem based on a dict of BytesIO objects
+
+    This is a global filesystem so instances of this class all point to the same
+    in memory filesystem.
+    """
+
+    store: ClassVar[dict[str, Any]] = {}  # global, do not overwrite!
+    pseudo_dirs = [""]  # global, do not overwrite!
+    protocol = "memory"
+    root_marker = "/"
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        if isinstance(path, PurePath):
+            if isinstance(path, PureWindowsPath):
+                return LocalFileSystem._strip_protocol(path)
+            else:
+                path = stringify_path(path)
+
+        path = path.removeprefix("memory://")
+        if "::" in path or "://" in path:
+            return path.rstrip("/")
+        path = path.lstrip("/").rstrip("/")
+        return "/" + path if path else ""
+
+    def ls(self, path, detail=True, **kwargs):
+        path = self._strip_protocol(path)
+        if path in self.store:
+            # there is a key with this exact name
+            if not detail:
+                return [path]
+            return [
+                {
+                    "name": path,
+                    "size": self.store[path].size,
+                    "type": "file",
+                    "created": self.store[path].created.timestamp(),
+                }
+            ]
+        paths = set()
+        starter = path + "/"
+        out = []
+        for p2 in tuple(self.store):
+            if p2.startswith(starter):
+                if "/" not in p2[len(starter) :]:
+                    # exact child
+                    out.append(
+                        {
+                            "name": p2,
+                            "size": self.store[p2].size,
+                            "type": "file",
+                            "created": self.store[p2].created.timestamp(),
+                        }
+                    )
+                elif len(p2) > len(starter):
+                    # implied child directory
+                    ppath = starter + p2[len(starter) :].split("/", 1)[0]
+                    if ppath not in paths:
+                        out = out or []
+                        out.append(
+                            {
+                                "name": ppath,
+                                "size": 0,
+                                "type": "directory",
+                            }
+                        )
+                        paths.add(ppath)
+        for p2 in self.pseudo_dirs:
+            if p2.startswith(starter):
+                if "/" not in p2[len(starter) :]:
+                    # exact child pdir
+                    if p2 not in paths:
+                        out.append({"name": p2, "size": 0, "type": "directory"})
+                        paths.add(p2)
+                else:
+                    # directory implied by deeper pdir
+                    ppath = starter + p2[len(starter) :].split("/", 1)[0]
+                    if ppath not in paths:
+                        out.append({"name": ppath, "size": 0, "type": "directory"})
+                        paths.add(ppath)
+        if not out:
+            if path in self.pseudo_dirs:
+                # empty dir
+                return []
+            raise FileNotFoundError(path)
+        if detail:
+            return out
+        return sorted([f["name"] for f in out])
+
+    def mkdir(self, path, create_parents=True, **kwargs):
+        path = self._strip_protocol(path)
+        if path in self.store or path in self.pseudo_dirs:
+            raise FileExistsError(path)
+        if self._parent(path).strip("/") and self.isfile(self._parent(path)):
+            raise NotADirectoryError(self._parent(path))
+        if create_parents and self._parent(path).strip("/"):
+            try:
+                self.mkdir(self._parent(path), create_parents, **kwargs)
+            except FileExistsError:
+                pass
+        if path and path not in self.pseudo_dirs:
+            self.pseudo_dirs.append(path)
+
+    def makedirs(self, path, exist_ok=False):
+        try:
+            self.mkdir(path, create_parents=True)
+        except FileExistsError:
+            if not exist_ok:
+                raise
+
+    def pipe_file(self, path, value, mode="overwrite", **kwargs):
+        """Set the bytes of given file
+
+        Avoids copies of the data if possible
+        """
+        mode = "xb" if mode == "create" else "wb"
+        self.open(path, mode=mode, data=value)
+
+    def rmdir(self, path):
+        path = self._strip_protocol(path)
+        if path == "":
+            # silently avoid deleting FS root
+            return
+        if path in self.pseudo_dirs:
+            if not self.ls(path):
+                self.pseudo_dirs.remove(path)
+            else:
+                raise OSError(ENOTEMPTY, "Directory not empty", path)
+        else:
+            raise FileNotFoundError(path)
+
+    def info(self, path, **kwargs):
+        logger.debug("info: %s", path)
+        path = self._strip_protocol(path)
+        if path in self.pseudo_dirs or any(
+            p.startswith(path + "/") for p in list(self.store) + self.pseudo_dirs
+        ):
+            return {
+                "name": path,
+                "size": 0,
+                "type": "directory",
+            }
+        elif path in self.store:
+            filelike = self.store[path]
+            return {
+                "name": path,
+                "size": filelike.size,
+                "type": "file",
+                "created": getattr(filelike, "created", None),
+            }
+        else:
+            raise FileNotFoundError(path)
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        autocommit=True,
+        cache_options=None,
+        **kwargs,
+    ):
+        path = self._strip_protocol(path)
+        if "x" in mode and self.exists(path):
+            raise FileExistsError
+        if path in self.pseudo_dirs:
+            raise IsADirectoryError(path)
+        parent = path
+        while len(parent) > 1:
+            parent = self._parent(parent)
+            if self.isfile(parent):
+                raise FileExistsError(parent)
+        if mode in ["rb", "ab", "r+b", "a+b"]:
+            if path in self.store:
+                f = self.store[path]
+                if "a" in mode:
+                    # position at the end of file
+                    f.seek(0, 2)
+                else:
+                    # position at the beginning of file
+                    f.seek(0)
+                return f
+            else:
+                raise FileNotFoundError(path)
+        elif mode in {"wb", "w+b", "xb", "x+b"}:
+            if "x" in mode and self.exists(path):
+                raise FileExistsError
+            m = MemoryFile(self, path, kwargs.get("data"))
+            if not self._intrans:
+                m.commit()
+            return m
+        else:
+            name = self.__class__.__name__
+            raise ValueError(f"unsupported file mode for {name}: {mode!r}")
+
+    def cp_file(self, path1, path2, **kwargs):
+        path1 = self._strip_protocol(path1)
+        path2 = self._strip_protocol(path2)
+        if self.isfile(path1):
+            self.store[path2] = MemoryFile(
+                self, path2, self.store[path1].getvalue()
+            )  # implicit copy
+        elif self.isdir(path1):
+            if path2 not in self.pseudo_dirs:
+                self.pseudo_dirs.append(path2)
+        else:
+            raise FileNotFoundError(path1)
+
+    def cat_file(self, path, start=None, end=None, **kwargs):
+        logger.debug("cat: %s", path)
+        path = self._strip_protocol(path)
+        try:
+            return bytes(self.store[path].getbuffer()[start:end])
+        except KeyError as e:
+            raise FileNotFoundError(path) from e
+
+    def _rm(self, path):
+        path = self._strip_protocol(path)
+        try:
+            del self.store[path]
+        except KeyError as e:
+            raise FileNotFoundError(path) from e
+
+    def modified(self, path):
+        path = self._strip_protocol(path)
+        try:
+            return self.store[path].modified
+        except KeyError as e:
+            raise FileNotFoundError(path) from e
+
+    def created(self, path):
+        path = self._strip_protocol(path)
+        try:
+            return self.store[path].created
+        except KeyError as e:
+            raise FileNotFoundError(path) from e
+
+    def isfile(self, path):
+        path = self._strip_protocol(path)
+        return path in self.store
+
+    def rm(self, path, recursive=False, maxdepth=None):
+        if isinstance(path, str):
+            path = self._strip_protocol(path)
+        else:
+            path = [self._strip_protocol(p) for p in path]
+        paths = self.expand_path(path, recursive=recursive, maxdepth=maxdepth)
+        for p in reversed(paths):
+            if self.isfile(p):
+                self.rm_file(p)
+            # If the expanded path doesn't exist, it is only because the expanded
+            # path was a directory that does not exist in self.pseudo_dirs. This
+            # is possible if you directly create files without making the
+            # directories first.
+            elif not self.exists(p):
+                continue
+            else:
+                self.rmdir(p)
+
+
+class MemoryFile(BytesIO):
+    """A BytesIO which can't close and works as a context manager
+
+    Can initialise with data. Each path should only be active once at any moment.
+
+    No need to provide fs, path if auto-committing (default)
+    """
+
+    def __init__(self, fs=None, path=None, data=None):
+        logger.debug("open file %s", path)
+        self.fs = fs
+        self.path = path
+        self.created = datetime.now(tz=timezone.utc)
+        self.modified = datetime.now(tz=timezone.utc)
+        if data:
+            super().__init__(data)
+            self.seek(0)
+
+    @property
+    def size(self):
+        return self.getbuffer().nbytes
+
+    def __enter__(self):
+        return self
+
+    def close(self):
+        pass
+
+    def discard(self):
+        pass
+
+    def commit(self):
+        self.fs.store[self.path] = self
+        self.modified = datetime.now(tz=timezone.utc)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/sftp.py

@@ -0,0 +1,187 @@
+import datetime
+import logging
+import os
+import types
+import uuid
+from stat import S_ISDIR, S_ISLNK
+
+import paramiko
+
+from .. import AbstractFileSystem
+from ..utils import infer_storage_options
+
+logger = logging.getLogger("fsspec.sftp")
+
+
+class SFTPFileSystem(AbstractFileSystem):
+    """Files over SFTP/SSH
+
+    Peer-to-peer filesystem over SSH using paramiko.
+
+    Note: if using this with the ``open`` or ``open_files``, with full URLs,
+    there is no way to tell if a path is relative, so all paths are assumed
+    to be absolute.
+    """
+
+    protocol = "sftp", "ssh"
+
+    def __init__(self, host, **ssh_kwargs):
+        """
+
+        Parameters
+        ----------
+        host: str
+            Hostname or IP as a string
+        temppath: str
+            Location on the server to put files, when within a transaction
+        ssh_kwargs: dict
+            Parameters passed on to connection. See details in
+            https://docs.paramiko.org/en/3.3/api/client.html#paramiko.client.SSHClient.connect
+            May include port, username, password...
+        """
+        if self._cached:
+            return
+        super().__init__(**ssh_kwargs)
+        self.temppath = ssh_kwargs.pop("temppath", "/tmp")  # remote temp directory
+        self.host = host
+        self.ssh_kwargs = ssh_kwargs
+        self._connect()
+
+    def _connect(self):
+        logger.debug("Connecting to SFTP server %s", self.host)
+        self.client = paramiko.SSHClient()
+        self.client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
+        self.client.connect(self.host, **self.ssh_kwargs)
+        self.ftp = self.client.open_sftp()
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        return infer_storage_options(path)["path"]
+
+    @staticmethod
+    def _get_kwargs_from_urls(urlpath):
+        out = infer_storage_options(urlpath)
+        out.pop("path", None)
+        out.pop("protocol", None)
+        return out
+
+    def mkdir(self, path, create_parents=True, mode=511):
+        path = self._strip_protocol(path)
+        logger.debug("Creating folder %s", path)
+        if self.exists(path):
+            raise FileExistsError(f"File exists: {path}")
+
+        if create_parents:
+            self.makedirs(path)
+        else:
+            self.ftp.mkdir(path, mode)
+
+    def makedirs(self, path, exist_ok=False, mode=511):
+        if self.exists(path) and not exist_ok:
+            raise FileExistsError(f"File exists: {path}")
+
+        parts = path.split("/")
+        new_path = "/" if path[:1] == "/" else ""
+
+        for part in parts:
+            if part:
+                new_path = f"{new_path}/{part}" if new_path else part
+                if not self.exists(new_path):
+                    self.ftp.mkdir(new_path, mode)
+
+    def rmdir(self, path):
+        path = self._strip_protocol(path)
+        logger.debug("Removing folder %s", path)
+        self.ftp.rmdir(path)
+
+    def info(self, path):
+        path = self._strip_protocol(path)
+        stat = self._decode_stat(self.ftp.stat(path))
+        stat["name"] = path
+        return stat
+
+    @staticmethod
+    def _decode_stat(stat, parent_path=None):
+        if S_ISDIR(stat.st_mode):
+            t = "directory"
+        elif S_ISLNK(stat.st_mode):
+            t = "link"
+        else:
+            t = "file"
+        out = {
+            "name": "",
+            "size": stat.st_size,
+            "type": t,
+            "uid": stat.st_uid,
+            "gid": stat.st_gid,
+            "time": datetime.datetime.fromtimestamp(
+                stat.st_atime, tz=datetime.timezone.utc
+            ),
+            "mtime": datetime.datetime.fromtimestamp(
+                stat.st_mtime, tz=datetime.timezone.utc
+            ),
+        }
+        if parent_path:
+            out["name"] = "/".join([parent_path.rstrip("/"), stat.filename])
+        return out
+
+    def ls(self, path, detail=False):
+        path = self._strip_protocol(path)
+        logger.debug("Listing folder %s", path)
+        stats = [self._decode_stat(stat, path) for stat in self.ftp.listdir_iter(path)]
+        if detail:
+            return stats
+        else:
+            paths = [stat["name"] for stat in stats]
+            return sorted(paths)
+
+    def put(self, lpath, rpath, callback=None, **kwargs):
+        rpath = self._strip_protocol(rpath)
+        logger.debug("Put file %s into %s", lpath, rpath)
+        self.ftp.put(lpath, rpath)
+
+    def get_file(self, rpath, lpath, **kwargs):
+        if self.isdir(rpath):
+            os.makedirs(lpath, exist_ok=True)
+        else:
+            self.ftp.get(self._strip_protocol(rpath), lpath)
+
+    def _open(self, path, mode="rb", block_size=None, **kwargs):
+        """
+        block_size: int or None
+            If 0, no buffering, if 1, line buffering, if >1, buffer that many
+            bytes, if None use default from paramiko.
+        """
+        logger.debug("Opening file %s", path)
+        if kwargs.get("autocommit", True) is False:
+            # writes to temporary file, move on commit
+            path2 = "/".join([self.temppath, str(uuid.uuid4())])
+            f = self.ftp.open(path2, mode, bufsize=block_size if block_size else -1)
+            f.temppath = path2
+            f.targetpath = path
+            f.fs = self
+            f.commit = types.MethodType(commit_a_file, f)
+            f.discard = types.MethodType(discard_a_file, f)
+        else:
+            f = self.ftp.open(path, mode, bufsize=block_size if block_size else -1)
+        return f
+
+    def _rm(self, path):
+        if self.isdir(path):
+            self.ftp.rmdir(path)
+        else:
+            self.ftp.remove(path)
+
+    def mv(self, old, new):
+        new = self._strip_protocol(new)
+        old = self._strip_protocol(old)
+        logger.debug("Renaming %s into %s", old, new)
+        self.ftp.posix_rename(old, new)
+
+
+def commit_a_file(self):
+    self.fs.mv(self.temppath, self.targetpath)
+
+
+def discard_a_file(self):
+    self.fs._rm(self.temppath)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/smb.py

@@ -0,0 +1,416 @@
+"""
+This module contains SMBFileSystem class responsible for handling access to
+Windows Samba network shares by using package smbprotocol
+"""
+
+import datetime
+import re
+import uuid
+from stat import S_ISDIR, S_ISLNK
+
+import smbclient
+import smbprotocol.exceptions
+
+from .. import AbstractFileSystem
+from ..utils import infer_storage_options
+
+# ! pylint: disable=bad-continuation
+
+
+class SMBFileSystem(AbstractFileSystem):
+    """Allow reading and writing to Windows and Samba network shares.
+
+    When using `fsspec.open()` for getting a file-like object the URI
+    should be specified as this format:
+    ``smb://workgroup;user:password@server:port/share/folder/file.csv``.
+
+    Example::
+
+        >>> import fsspec
+        >>> with fsspec.open(
+        ...     'smb://myuser:mypassword@myserver.com/' 'share/folder/file.csv'
+        ... ) as smbfile:
+        ...     df = pd.read_csv(smbfile, sep='|', header=None)
+
+    Note that you need to pass in a valid hostname or IP address for the host
+    component of the URL. Do not use the Windows/NetBIOS machine name for the
+    host component.
+
+    The first component of the path in the URL points to the name of the shared
+    folder. Subsequent path components will point to the directory/folder/file.
+
+    The URL components ``workgroup`` , ``user``, ``password`` and ``port`` may be
+    optional.
+
+    .. note::
+
+        For working this source require `smbprotocol`_ to be installed, e.g.::
+
+            $ pip install smbprotocol
+            # or
+            # pip install smbprotocol[kerberos]
+
+    .. _smbprotocol: https://github.com/jborean93/smbprotocol#requirements
+
+    Note: if using this with the ``open`` or ``open_files``, with full URLs,
+    there is no way to tell if a path is relative, so all paths are assumed
+    to be absolute.
+    """
+
+    protocol = "smb"
+
+    # pylint: disable=too-many-arguments
+    def __init__(
+        self,
+        host,
+        port=None,
+        username=None,
+        password=None,
+        timeout=60,
+        encrypt=None,
+        share_access=None,
+        register_session_retries=4,
+        register_session_retry_wait=1,
+        register_session_retry_factor=10,
+        auto_mkdir=False,
+        **kwargs,
+    ):
+        """
+        You can use _get_kwargs_from_urls to get some kwargs from
+        a reasonable SMB url.
+
+        Authentication will be anonymous or integrated if username/password are not
+        given.
+
+        Parameters
+        ----------
+        host: str
+            The remote server name/ip to connect to
+        port: int or None
+            Port to connect with. Usually 445, sometimes 139.
+        username: str or None
+            Username to connect with. Required if Kerberos auth is not being used.
+        password: str or None
+            User's password on the server, if using username
+        timeout: int
+            Connection timeout in seconds
+        encrypt: bool
+            Whether to force encryption or not, once this has been set to True
+            the session cannot be changed back to False.
+        share_access: str or None
+            Specifies the default access applied to file open operations
+            performed with this file system object.
+            This affects whether other processes can concurrently open a handle
+            to the same file.
+
+            - None (the default): exclusively locks the file until closed.
+            - 'r': Allow other handles to be opened with read access.
+            - 'w': Allow other handles to be opened with write access.
+            - 'd': Allow other handles to be opened with delete access.
+        register_session_retries: int
+            Number of retries to register a session with the server. Retries are not performed
+            for authentication errors, as they are considered as invalid credentials and not network
+            issues. If set to negative value, no register attempts will be performed.
+        register_session_retry_wait: int
+            Time in seconds to wait between each retry. Number must be non-negative.
+        register_session_retry_factor: int
+            Base factor for the wait time between each retry. The wait time
+            is calculated using exponential function. For factor=1 all wait times
+            will be equal to `register_session_retry_wait`. For any number of retries,
+            the last wait time will be equal to `register_session_retry_wait` and for retries>1
+            the first wait time will be equal to `register_session_retry_wait / factor`.
+            Number must be equal to or greater than 1. Optimal factor is 10.
+        auto_mkdir: bool
+            Whether, when opening a file, the directory containing it should
+            be created (if it doesn't already exist). This is assumed by pyarrow
+            and zarr-python code.
+        """
+        super().__init__(**kwargs)
+        self.host = host
+        self.port = port
+        self.username = username
+        self.password = password
+        self.timeout = timeout
+        self.encrypt = encrypt
+        self.temppath = kwargs.pop("temppath", "")
+        self.share_access = share_access
+        self.register_session_retries = register_session_retries
+        if register_session_retry_wait < 0:
+            raise ValueError(
+                "register_session_retry_wait must be a non-negative integer"
+            )
+        self.register_session_retry_wait = register_session_retry_wait
+        if register_session_retry_factor < 1:
+            raise ValueError(
+                "register_session_retry_factor must be a positive "
+                "integer equal to or greater than 1"
+            )
+        self.register_session_retry_factor = register_session_retry_factor
+        self.auto_mkdir = auto_mkdir
+        self._connect()
+
+    @property
+    def _port(self):
+        return 445 if self.port is None else self.port
+
+    def _connect(self):
+        import time
+
+        if self.register_session_retries <= -1:
+            return
+
+        retried_errors = []
+
+        wait_time = self.register_session_retry_wait
+        n_waits = (
+            self.register_session_retries - 1
+        )  # -1 = No wait time after the last retry
+        factor = self.register_session_retry_factor
+
+        # Generate wait times for each retry attempt.
+        # Wait times are calculated using exponential function. For factor=1 all wait times
+        # will be equal to `wait`. For any number of retries the last wait time will be
+        # equal to `wait` and for retries>2 the first wait time will be equal to `wait / factor`.
+        wait_times = iter(
+            factor ** (n / n_waits - 1) * wait_time for n in range(0, n_waits + 1)
+        )
+
+        for attempt in range(self.register_session_retries + 1):
+            try:
+                smbclient.register_session(
+                    self.host,
+                    username=self.username,
+                    password=self.password,
+                    port=self._port,
+                    encrypt=self.encrypt,
+                    connection_timeout=self.timeout,
+                )
+                return
+            except (
+                smbprotocol.exceptions.SMBAuthenticationError,
+                smbprotocol.exceptions.LogonFailure,
+            ):
+                # These exceptions should not be repeated, as they clearly indicate
+                # that the credentials are invalid and not a network issue.
+                raise
+            except ValueError as exc:
+                if re.findall(r"\[Errno -\d+]", str(exc)):
+                    # This exception is raised by the smbprotocol.transport:Tcp.connect
+                    # and originates from socket.gaierror (OSError). These exceptions might
+                    # be raised due to network instability. We will retry to connect.
+                    retried_errors.append(exc)
+                else:
+                    # All another ValueError exceptions should be raised, as they are not
+                    # related to network issues.
+                    raise
+            except Exception as exc:
+                # Save the exception and retry to connect. This except might be dropped
+                # in the future, once all exceptions suited for retry are identified.
+                retried_errors.append(exc)
+
+            if attempt < self.register_session_retries:
+                time.sleep(next(wait_times))
+
+        # Raise last exception to inform user about the connection issues.
+        # Note: Should we use ExceptionGroup to raise all exceptions?
+        raise retried_errors[-1]
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        return infer_storage_options(path)["path"]
+
+    @staticmethod
+    def _get_kwargs_from_urls(path):
+        # smb://workgroup;user:password@host:port/share/folder/file.csv
+        out = infer_storage_options(path)
+        out.pop("path", None)
+        out.pop("protocol", None)
+        return out
+
+    def mkdir(self, path, create_parents=True, **kwargs):
+        wpath = _as_unc_path(self.host, path)
+        if create_parents:
+            smbclient.makedirs(wpath, exist_ok=False, port=self._port, **kwargs)
+        else:
+            smbclient.mkdir(wpath, port=self._port, **kwargs)
+
+    def makedirs(self, path, exist_ok=False):
+        if _share_has_path(path):
+            wpath = _as_unc_path(self.host, path)
+            smbclient.makedirs(wpath, exist_ok=exist_ok, port=self._port)
+
+    def rmdir(self, path):
+        if _share_has_path(path):
+            wpath = _as_unc_path(self.host, path)
+            smbclient.rmdir(wpath, port=self._port)
+
+    def info(self, path, **kwargs):
+        wpath = _as_unc_path(self.host, path)
+        stats = smbclient.stat(wpath, port=self._port, **kwargs)
+        if S_ISDIR(stats.st_mode):
+            stype = "directory"
+        elif S_ISLNK(stats.st_mode):
+            stype = "link"
+        else:
+            stype = "file"
+        res = {
+            "name": path + "/" if stype == "directory" else path,
+            "size": stats.st_size,
+            "type": stype,
+            "uid": stats.st_uid,
+            "gid": stats.st_gid,
+            "time": stats.st_atime,
+            "mtime": stats.st_mtime,
+        }
+        return res
+
+    def created(self, path):
+        """Return the created timestamp of a file as a datetime.datetime"""
+        wpath = _as_unc_path(self.host, path)
+        stats = smbclient.stat(wpath, port=self._port)
+        return datetime.datetime.fromtimestamp(stats.st_ctime, tz=datetime.timezone.utc)
+
+    def modified(self, path):
+        """Return the modified timestamp of a file as a datetime.datetime"""
+        wpath = _as_unc_path(self.host, path)
+        stats = smbclient.stat(wpath, port=self._port)
+        return datetime.datetime.fromtimestamp(stats.st_mtime, tz=datetime.timezone.utc)
+
+    def ls(self, path, detail=True, **kwargs):
+        unc = _as_unc_path(self.host, path)
+        listed = smbclient.listdir(unc, port=self._port, **kwargs)
+        dirs = ["/".join([path.rstrip("/"), p]) for p in listed]
+        if detail:
+            dirs = [self.info(d) for d in dirs]
+        return dirs
+
+    # pylint: disable=too-many-arguments
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=-1,
+        autocommit=True,
+        cache_options=None,
+        **kwargs,
+    ):
+        """
+        block_size: int or None
+            If 0, no buffering, 1, line buffering, >1, buffer that many bytes
+
+        Notes
+        -----
+        By specifying 'share_access' in 'kwargs' it is possible to override the
+        default shared access setting applied in the constructor of this object.
+        """
+        if self.auto_mkdir and "w" in mode:
+            self.makedirs(self._parent(path), exist_ok=True)
+        bls = block_size if block_size is not None and block_size >= 0 else -1
+        wpath = _as_unc_path(self.host, path)
+        share_access = kwargs.pop("share_access", self.share_access)
+        if "w" in mode and autocommit is False:
+            temp = _as_temp_path(self.host, path, self.temppath)
+            return SMBFileOpener(
+                wpath, temp, mode, port=self._port, block_size=bls, **kwargs
+            )
+        return smbclient.open_file(
+            wpath,
+            mode,
+            buffering=bls,
+            share_access=share_access,
+            port=self._port,
+            **kwargs,
+        )
+
+    def copy(self, path1, path2, **kwargs):
+        """Copy within two locations in the same filesystem"""
+        wpath1 = _as_unc_path(self.host, path1)
+        wpath2 = _as_unc_path(self.host, path2)
+        if self.auto_mkdir:
+            self.makedirs(self._parent(path2), exist_ok=True)
+        smbclient.copyfile(wpath1, wpath2, port=self._port, **kwargs)
+
+    def _rm(self, path):
+        if _share_has_path(path):
+            wpath = _as_unc_path(self.host, path)
+            stats = smbclient.stat(wpath, port=self._port)
+            if S_ISDIR(stats.st_mode):
+                smbclient.rmdir(wpath, port=self._port)
+            else:
+                smbclient.remove(wpath, port=self._port)
+
+    def mv(self, path1, path2, recursive=None, maxdepth=None, **kwargs):
+        wpath1 = _as_unc_path(self.host, path1)
+        wpath2 = _as_unc_path(self.host, path2)
+        smbclient.rename(wpath1, wpath2, port=self._port, **kwargs)
+
+
+def _as_unc_path(host, path):
+    rpath = path.replace("/", "\\")
+    unc = f"\\\\{host}{rpath}"
+    return unc
+
+
+def _as_temp_path(host, path, temppath):
+    share = path.split("/")[1]
+    temp_file = f"/{share}{temppath}/{uuid.uuid4()}"
+    unc = _as_unc_path(host, temp_file)
+    return unc
+
+
+def _share_has_path(path):
+    parts = path.count("/")
+    if path.endswith("/"):
+        return parts > 2
+    return parts > 1
+
+
+class SMBFileOpener:
+    """writes to remote temporary file, move on commit"""
+
+    def __init__(self, path, temp, mode, port=445, block_size=-1, **kwargs):
+        self.path = path
+        self.temp = temp
+        self.mode = mode
+        self.block_size = block_size
+        self.kwargs = kwargs
+        self.smbfile = None
+        self._incontext = False
+        self.port = port
+        self._open()
+
+    def _open(self):
+        if self.smbfile is None or self.smbfile.closed:
+            self.smbfile = smbclient.open_file(
+                self.temp,
+                self.mode,
+                port=self.port,
+                buffering=self.block_size,
+                **self.kwargs,
+            )
+
+    def commit(self):
+        """Move temp file to definitive on success."""
+        # TODO: use transaction support in SMB protocol
+        smbclient.replace(self.temp, self.path, port=self.port)
+
+    def discard(self):
+        """Remove the temp file on failure."""
+        smbclient.remove(self.temp, port=self.port)
+
+    def __fspath__(self):
+        return self.path
+
+    def __iter__(self):
+        return self.smbfile.__iter__()
+
+    def __getattr__(self, item):
+        return getattr(self.smbfile, item)
+
+    def __enter__(self):
+        self._incontext = True
+        return self.smbfile.__enter__()
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        self._incontext = False
+        self.smbfile.__exit__(exc_type, exc_value, traceback)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/tar.py

@@ -0,0 +1,124 @@
+import logging
+import tarfile
+
+import fsspec
+from fsspec.archive import AbstractArchiveFileSystem
+from fsspec.compression import compr
+from fsspec.utils import infer_compression
+
+typemap = {b"0": "file", b"5": "directory"}
+
+logger = logging.getLogger("tar")
+
+
+class TarFileSystem(AbstractArchiveFileSystem):
+    """Compressed Tar archives as a file-system (read-only)
+
+    Supports the following formats:
+    tar.gz, tar.bz2, tar.xz
+    """
+
+    root_marker = ""
+    protocol = "tar"
+    cachable = False
+
+    def __init__(
+        self,
+        fo="",
+        index_store=None,
+        target_options=None,
+        target_protocol=None,
+        compression=None,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        target_options = target_options or {}
+
+        if isinstance(fo, str):
+            self.of = fsspec.open(fo, protocol=target_protocol, **target_options)
+            fo = self.of.open()  # keep the reference
+
+        # Try to infer compression.
+        if compression is None:
+            name = None
+
+            # Try different ways to get hold of the filename. `fo` might either
+            # be a `fsspec.LocalFileOpener`, an `io.BufferedReader` or an
+            # `fsspec.AbstractFileSystem` instance.
+            try:
+                # Amended io.BufferedReader or similar.
+                # This uses a "protocol extension" where original filenames are
+                # propagated to archive-like filesystems in order to let them
+                # infer the right compression appropriately.
+                if hasattr(fo, "original"):
+                    name = fo.original
+
+                # fsspec.LocalFileOpener
+                elif hasattr(fo, "path"):
+                    name = fo.path
+
+                # io.BufferedReader
+                elif hasattr(fo, "name"):
+                    name = fo.name
+
+                # fsspec.AbstractFileSystem
+                elif hasattr(fo, "info"):
+                    name = fo.info()["name"]
+
+            except Exception as ex:
+                logger.warning(
+                    f"Unable to determine file name, not inferring compression: {ex}"
+                )
+
+            if name is not None:
+                compression = infer_compression(name)
+                logger.info(f"Inferred compression {compression} from file name {name}")
+
+        if compression is not None:
+            # TODO: tarfile already implements compression with modes like "'r:gz'",
+            #  but then would seek to offset in the file work?
+            fo = compr[compression](fo)
+
+        self._fo_ref = fo
+        self.fo = fo  # the whole instance is a context
+        self.tar = tarfile.TarFile(fileobj=self.fo)
+        self.dir_cache = None
+
+        self.index_store = index_store
+        self.index = None
+        self._index()
+
+    def _index(self):
+        # TODO: load and set saved index, if exists
+        out = {}
+        for ti in self.tar:
+            info = ti.get_info()
+            info["type"] = typemap.get(info["type"], "file")
+            name = ti.get_info()["name"].rstrip("/")
+            out[name] = (info, ti.offset_data)
+
+        self.index = out
+        # TODO: save index to self.index_store here, if set
+
+    def _get_dirs(self):
+        if self.dir_cache is not None:
+            return
+
+        # This enables ls to get directories as children as well as files
+        self.dir_cache = {
+            dirname: {"name": dirname, "size": 0, "type": "directory"}
+            for dirname in self._all_dirnames(self.tar.getnames())
+        }
+        for member in self.tar.getmembers():
+            info = member.get_info()
+            info["name"] = info["name"].rstrip("/")
+            info["type"] = typemap.get(info["type"], "file")
+            self.dir_cache[info["name"]] = info
+
+    def _open(self, path, mode="rb", **kwargs):
+        if mode != "rb":
+            raise ValueError("Read-only filesystem implementation")
+        details, offset = self.index[path]
+        if details["type"] != "file":
+            raise ValueError("Can only handle regular files")
+        return self.tar.extractfile(path)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/webhdfs.py

@@ -0,0 +1,485 @@
+# https://hadoop.apache.org/docs/r1.0.4/webhdfs.html
+
+import logging
+import os
+import secrets
+import shutil
+import tempfile
+import uuid
+from contextlib import suppress
+from urllib.parse import quote
+
+import requests
+
+from ..spec import AbstractBufferedFile, AbstractFileSystem
+from ..utils import infer_storage_options, tokenize
+
+logger = logging.getLogger("webhdfs")
+
+
+class WebHDFS(AbstractFileSystem):
+    """
+    Interface to HDFS over HTTP using the WebHDFS API. Supports also HttpFS gateways.
+
+    Four auth mechanisms are supported:
+
+    insecure: no auth is done, and the user is assumed to be whoever they
+        say they are (parameter ``user``), or a predefined value such as
+        "dr.who" if not given
+    spnego: when kerberos authentication is enabled, auth is negotiated by
+        requests_kerberos https://github.com/requests/requests-kerberos .
+        This establishes a session based on existing kinit login and/or
+        specified principal/password; parameters are passed with ``kerb_kwargs``
+    token: uses an existing Hadoop delegation token from another secured
+        service. Indeed, this client can also generate such tokens when
+        not insecure. Note that tokens expire, but can be renewed (by a
+        previously specified user) and may allow for proxying.
+    basic-auth: used when both parameter ``user`` and parameter ``password``
+        are provided.
+
+    """
+
+    tempdir = str(tempfile.gettempdir())
+    protocol = "webhdfs", "webHDFS"
+
+    def __init__(
+        self,
+        host,
+        port=50070,
+        kerberos=False,
+        token=None,
+        user=None,
+        password=None,
+        proxy_to=None,
+        kerb_kwargs=None,
+        data_proxy=None,
+        use_https=False,
+        session_cert=None,
+        session_verify=True,
+        **kwargs,
+    ):
+        """
+        Parameters
+        ----------
+        host: str
+            Name-node address
+        port: int
+            Port for webHDFS
+        kerberos: bool
+            Whether to authenticate with kerberos for this connection
+        token: str or None
+            If given, use this token on every call to authenticate. A user
+            and user-proxy may be encoded in the token and should not be also
+            given
+        user: str or None
+            If given, assert the user name to connect with
+        password: str or None
+            If given, assert the password to use for basic auth. If password
+            is provided, user must be provided also
+        proxy_to: str or None
+            If given, the user has the authority to proxy, and this value is
+            the user in who's name actions are taken
+        kerb_kwargs: dict
+            Any extra arguments for HTTPKerberosAuth, see
+            `<https://github.com/requests/requests-kerberos/blob/master/requests_kerberos/kerberos_.py>`_
+        data_proxy: dict, callable or None
+            If given, map data-node addresses. This can be necessary if the
+            HDFS cluster is behind a proxy, running on Docker or otherwise has
+            a mismatch between the host-names given by the name-node and the
+            address by which to refer to them from the client. If a dict,
+            maps host names ``host->data_proxy[host]``; if a callable, full
+            URLs are passed, and function must conform to
+            ``url->data_proxy(url)``.
+        use_https: bool
+            Whether to connect to the Name-node using HTTPS instead of HTTP
+        session_cert: str or Tuple[str, str] or None
+            Path to a certificate file, or tuple of (cert, key) files to use
+            for the requests.Session
+        session_verify: str, bool or None
+            Path to a certificate file to use for verifying the requests.Session.
+        kwargs
+        """
+        if self._cached:
+            return
+        super().__init__(**kwargs)
+        self.url = f"{'https' if use_https else 'http'}://{host}:{port}/webhdfs/v1"
+        self.kerb = kerberos
+        self.kerb_kwargs = kerb_kwargs or {}
+        self.pars = {}
+        self.proxy = data_proxy or {}
+        if token is not None:
+            if user is not None or proxy_to is not None:
+                raise ValueError(
+                    "If passing a delegation token, must not set "
+                    "user or proxy_to, as these are encoded in the"
+                    " token"
+                )
+            self.pars["delegation"] = token
+        self.user = user
+        self.password = password
+
+        if password is not None:
+            if user is None:
+                raise ValueError(
+                    "If passing a password, the user must also be"
+                    "set in order to set up the basic-auth"
+                )
+        else:
+            if user is not None:
+                self.pars["user.name"] = user
+
+        if proxy_to is not None:
+            self.pars["doas"] = proxy_to
+        if kerberos and user is not None:
+            raise ValueError(
+                "If using Kerberos auth, do not specify the "
+                "user, this is handled by kinit."
+            )
+
+        self.session_cert = session_cert
+        self.session_verify = session_verify
+
+        self._connect()
+
+        self._fsid = f"webhdfs_{tokenize(host, port)}"
+
+    @property
+    def fsid(self):
+        return self._fsid
+
+    def _connect(self):
+        self.session = requests.Session()
+
+        if self.session_cert:
+            self.session.cert = self.session_cert
+
+        self.session.verify = self.session_verify
+
+        if self.kerb:
+            from requests_kerberos import HTTPKerberosAuth
+
+            self.session.auth = HTTPKerberosAuth(**self.kerb_kwargs)
+
+        if self.user is not None and self.password is not None:
+            from requests.auth import HTTPBasicAuth
+
+            self.session.auth = HTTPBasicAuth(self.user, self.password)
+
+    def _call(self, op, method="get", path=None, data=None, redirect=True, **kwargs):
+        path = self._strip_protocol(path) if path is not None else ""
+        url = self._apply_proxy(self.url + quote(path, safe="/="))
+        args = kwargs.copy()
+        args.update(self.pars)
+        args["op"] = op.upper()
+        logger.debug("sending %s with %s", url, method)
+        out = self.session.request(
+            method=method.upper(),
+            url=url,
+            params=args,
+            data=data,
+            allow_redirects=redirect,
+        )
+        if out.status_code in [400, 401, 403, 404, 500]:
+            try:
+                err = out.json()
+                msg = err["RemoteException"]["message"]
+                exp = err["RemoteException"]["exception"]
+            except (ValueError, KeyError):
+                pass
+            else:
+                if exp in ["IllegalArgumentException", "UnsupportedOperationException"]:
+                    raise ValueError(msg)
+                elif exp in ["SecurityException", "AccessControlException"]:
+                    raise PermissionError(msg)
+                elif exp in ["FileNotFoundException"]:
+                    raise FileNotFoundError(msg)
+                else:
+                    raise RuntimeError(msg)
+        out.raise_for_status()
+        return out
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        autocommit=True,
+        replication=None,
+        permissions=None,
+        **kwargs,
+    ):
+        """
+
+        Parameters
+        ----------
+        path: str
+            File location
+        mode: str
+            'rb', 'wb', etc.
+        block_size: int
+            Client buffer size for read-ahead or write buffer
+        autocommit: bool
+            If False, writes to temporary file that only gets put in final
+            location upon commit
+        replication: int
+            Number of copies of file on the cluster, write mode only
+        permissions: str or int
+            posix permissions, write mode only
+        kwargs
+
+        Returns
+        -------
+        WebHDFile instance
+        """
+        block_size = block_size or self.blocksize
+        return WebHDFile(
+            self,
+            path,
+            mode=mode,
+            block_size=block_size,
+            tempdir=self.tempdir,
+            autocommit=autocommit,
+            replication=replication,
+            permissions=permissions,
+        )
+
+    @staticmethod
+    def _process_info(info):
+        info["type"] = info["type"].lower()
+        info["size"] = info["length"]
+        return info
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        return infer_storage_options(path)["path"]
+
+    @staticmethod
+    def _get_kwargs_from_urls(urlpath):
+        out = infer_storage_options(urlpath)
+        out.pop("path", None)
+        out.pop("protocol", None)
+        if "username" in out:
+            out["user"] = out.pop("username")
+        return out
+
+    def info(self, path):
+        out = self._call("GETFILESTATUS", path=path)
+        info = out.json()["FileStatus"]
+        info["name"] = path
+        return self._process_info(info)
+
+    def ls(self, path, detail=False, **kwargs):
+        out = self._call("LISTSTATUS", path=path)
+        infos = out.json()["FileStatuses"]["FileStatus"]
+        for info in infos:
+            self._process_info(info)
+            info["name"] = path.rstrip("/") + "/" + info["pathSuffix"]
+        if detail:
+            return sorted(infos, key=lambda i: i["name"])
+        else:
+            return sorted(info["name"] for info in infos)
+
+    def content_summary(self, path):
+        """Total numbers of files, directories and bytes under path"""
+        out = self._call("GETCONTENTSUMMARY", path=path)
+        return out.json()["ContentSummary"]
+
+    def ukey(self, path):
+        """Checksum info of file, giving method and result"""
+        out = self._call("GETFILECHECKSUM", path=path, redirect=False)
+        if "Location" in out.headers:
+            location = self._apply_proxy(out.headers["Location"])
+            out2 = self.session.get(location)
+            out2.raise_for_status()
+            return out2.json()["FileChecksum"]
+        else:
+            out.raise_for_status()
+            return out.json()["FileChecksum"]
+
+    def home_directory(self):
+        """Get user's home directory"""
+        out = self._call("GETHOMEDIRECTORY")
+        return out.json()["Path"]
+
+    def get_delegation_token(self, renewer=None):
+        """Retrieve token which can give the same authority to other uses
+
+        Parameters
+        ----------
+        renewer: str or None
+            User who may use this token; if None, will be current user
+        """
+        if renewer:
+            out = self._call("GETDELEGATIONTOKEN", renewer=renewer)
+        else:
+            out = self._call("GETDELEGATIONTOKEN")
+        t = out.json()["Token"]
+        if t is None:
+            raise ValueError("No token available for this user/security context")
+        return t["urlString"]
+
+    def renew_delegation_token(self, token):
+        """Make token live longer. Returns new expiry time"""
+        out = self._call("RENEWDELEGATIONTOKEN", method="put", token=token)
+        return out.json()["long"]
+
+    def cancel_delegation_token(self, token):
+        """Stop the token from being useful"""
+        self._call("CANCELDELEGATIONTOKEN", method="put", token=token)
+
+    def chmod(self, path, mod):
+        """Set the permission at path
+
+        Parameters
+        ----------
+        path: str
+            location to set (file or directory)
+        mod: str or int
+            posix epresentation or permission, give as oct string, e.g, '777'
+            or 0o777
+        """
+        self._call("SETPERMISSION", method="put", path=path, permission=mod)
+
+    def chown(self, path, owner=None, group=None):
+        """Change owning user and/or group"""
+        kwargs = {}
+        if owner is not None:
+            kwargs["owner"] = owner
+        if group is not None:
+            kwargs["group"] = group
+        self._call("SETOWNER", method="put", path=path, **kwargs)
+
+    def set_replication(self, path, replication):
+        """
+        Set file replication factor
+
+        Parameters
+        ----------
+        path: str
+            File location (not for directories)
+        replication: int
+            Number of copies of file on the cluster. Should be smaller than
+            number of data nodes; normally 3 on most systems.
+        """
+        self._call("SETREPLICATION", path=path, method="put", replication=replication)
+
+    def mkdir(self, path, **kwargs):
+        self._call("MKDIRS", method="put", path=path)
+
+    def makedirs(self, path, exist_ok=False):
+        if exist_ok is False and self.exists(path):
+            raise FileExistsError(path)
+        self.mkdir(path)
+
+    def mv(self, path1, path2, **kwargs):
+        self._call("RENAME", method="put", path=path1, destination=path2)
+
+    def rm(self, path, recursive=False, **kwargs):
+        self._call(
+            "DELETE",
+            method="delete",
+            path=path,
+            recursive="true" if recursive else "false",
+        )
+
+    def rm_file(self, path, **kwargs):
+        self.rm(path)
+
+    def cp_file(self, lpath, rpath, **kwargs):
+        with self.open(lpath) as lstream:
+            tmp_fname = "/".join([self._parent(rpath), f".tmp.{secrets.token_hex(16)}"])
+            # Perform an atomic copy (stream to a temporary file and
+            # move it to the actual destination).
+            try:
+                with self.open(tmp_fname, "wb") as rstream:
+                    shutil.copyfileobj(lstream, rstream)
+                self.mv(tmp_fname, rpath)
+            except BaseException:
+                with suppress(FileNotFoundError):
+                    self.rm(tmp_fname)
+                raise
+
+    def _apply_proxy(self, location):
+        if self.proxy and callable(self.proxy):
+            location = self.proxy(location)
+        elif self.proxy:
+            # as a dict
+            for k, v in self.proxy.items():
+                location = location.replace(k, v, 1)
+        return location
+
+
+class WebHDFile(AbstractBufferedFile):
+    """A file living in HDFS over webHDFS"""
+
+    def __init__(self, fs, path, **kwargs):
+        super().__init__(fs, path, **kwargs)
+        kwargs = kwargs.copy()
+        if kwargs.get("permissions", None) is None:
+            kwargs.pop("permissions", None)
+        if kwargs.get("replication", None) is None:
+            kwargs.pop("replication", None)
+        self.permissions = kwargs.pop("permissions", 511)
+        tempdir = kwargs.pop("tempdir")
+        if kwargs.pop("autocommit", False) is False:
+            self.target = self.path
+            self.path = os.path.join(tempdir, str(uuid.uuid4()))
+
+    def _upload_chunk(self, final=False):
+        """Write one part of a multi-block file upload
+
+        Parameters
+        ==========
+        final: bool
+            This is the last block, so should complete file, if
+            self.autocommit is True.
+        """
+        out = self.fs.session.post(
+            self.location,
+            data=self.buffer.getvalue(),
+            headers={"content-type": "application/octet-stream"},
+        )
+        out.raise_for_status()
+        return True
+
+    def _initiate_upload(self):
+        """Create remote file/upload"""
+        kwargs = self.kwargs.copy()
+        if "a" in self.mode:
+            op, method = "APPEND", "POST"
+        else:
+            op, method = "CREATE", "PUT"
+            kwargs["overwrite"] = "true"
+        out = self.fs._call(op, method, self.path, redirect=False, **kwargs)
+        location = self.fs._apply_proxy(out.headers["Location"])
+        if "w" in self.mode:
+            # create empty file to append to
+            out2 = self.fs.session.put(
+                location, headers={"content-type": "application/octet-stream"}
+            )
+            out2.raise_for_status()
+            # after creating empty file, change location to append to
+            out2 = self.fs._call("APPEND", "POST", self.path, redirect=False, **kwargs)
+            self.location = self.fs._apply_proxy(out2.headers["Location"])
+
+    def _fetch_range(self, start, end):
+        start = max(start, 0)
+        end = min(self.size, end)
+        if start >= end or start >= self.size:
+            return b""
+        out = self.fs._call(
+            "OPEN", path=self.path, offset=start, length=end - start, redirect=False
+        )
+        out.raise_for_status()
+        if "Location" in out.headers:
+            location = out.headers["Location"]
+            out2 = self.fs.session.get(self.fs._apply_proxy(location))
+            return out2.content
+        else:
+            return out.content
+
+    def commit(self):
+        self.fs.mv(self.path, self.target)
+
+    def discard(self):
+        self.fs.rm(self.path)

backend_service/venv/lib/python3.13/site-packages/fsspec/implementations/zip.py

@@ -0,0 +1,177 @@
+import os
+import zipfile
+
+import fsspec
+from fsspec.archive import AbstractArchiveFileSystem
+
+
+class ZipFileSystem(AbstractArchiveFileSystem):
+    """Read/Write contents of ZIP archive as a file-system
+
+    Keeps file object open while instance lives.
+
+    This class is pickleable, but not necessarily thread-safe
+    """
+
+    root_marker = ""
+    protocol = "zip"
+    cachable = False
+
+    def __init__(
+        self,
+        fo="",
+        mode="r",
+        target_protocol=None,
+        target_options=None,
+        compression=zipfile.ZIP_STORED,
+        allowZip64=True,
+        compresslevel=None,
+        **kwargs,
+    ):
+        """
+        Parameters
+        ----------
+        fo: str or file-like
+            Contains ZIP, and must exist. If a str, will fetch file using
+            :meth:`~fsspec.open_files`, which must return one file exactly.
+        mode: str
+            Accept: "r", "w", "a"
+        target_protocol: str (optional)
+            If ``fo`` is a string, this value can be used to override the
+            FS protocol inferred from a URL
+        target_options: dict (optional)
+            Kwargs passed when instantiating the target FS, if ``fo`` is
+            a string.
+        compression, allowZip64, compresslevel: passed to ZipFile
+            Only relevant when creating a ZIP
+        """
+        super().__init__(self, **kwargs)
+        if mode not in set("rwa"):
+            raise ValueError(f"mode '{mode}' no understood")
+        self.mode = mode
+        if isinstance(fo, (str, os.PathLike)):
+            if mode == "a":
+                m = "r+b"
+            else:
+                m = mode + "b"
+            fo = fsspec.open(
+                fo, mode=m, protocol=target_protocol, **(target_options or {})
+            )
+        self.force_zip_64 = allowZip64
+        self.of = fo
+        self.fo = fo.__enter__()  # the whole instance is a context
+        self.zip = zipfile.ZipFile(
+            self.fo,
+            mode=mode,
+            compression=compression,
+            allowZip64=allowZip64,
+            compresslevel=compresslevel,
+        )
+        self.dir_cache = None
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        # zip file paths are always relative to the archive root
+        return super()._strip_protocol(path).lstrip("/")
+
+    def __del__(self):
+        if hasattr(self, "zip"):
+            self.close()
+            del self.zip
+
+    def close(self):
+        """Commits any write changes to the file. Done on ``del`` too."""
+        self.zip.close()
+
+    def _get_dirs(self):
+        if self.dir_cache is None or self.mode in set("wa"):
+            # when writing, dir_cache is always in the ZipFile's attributes,
+            # not read from the file.
+            files = self.zip.infolist()
+            self.dir_cache = {
+                dirname.rstrip("/"): {
+                    "name": dirname.rstrip("/"),
+                    "size": 0,
+                    "type": "directory",
+                }
+                for dirname in self._all_dirnames(self.zip.namelist())
+            }
+            for z in files:
+                f = {s: getattr(z, s, None) for s in zipfile.ZipInfo.__slots__}
+                f.update(
+                    {
+                        "name": z.filename.rstrip("/"),
+                        "size": z.file_size,
+                        "type": ("directory" if z.is_dir() else "file"),
+                    }
+                )
+                self.dir_cache[f["name"]] = f
+
+    def pipe_file(self, path, value, **kwargs):
+        # override upstream, because we know the exact file size in this case
+        self.zip.writestr(path, value, **kwargs)
+
+    def _open(
+        self,
+        path,
+        mode="rb",
+        block_size=None,
+        autocommit=True,
+        cache_options=None,
+        **kwargs,
+    ):
+        path = self._strip_protocol(path)
+        if "r" in mode and self.mode in set("wa"):
+            if self.exists(path):
+                raise OSError("ZipFS can only be open for reading or writing, not both")
+            raise FileNotFoundError(path)
+        if "r" in self.mode and "w" in mode:
+            raise OSError("ZipFS can only be open for reading or writing, not both")
+        out = self.zip.open(path, mode.strip("b"), force_zip64=self.force_zip_64)
+        if "r" in mode:
+            info = self.info(path)
+            out.size = info["size"]
+            out.name = info["name"]
+        return out
+
+    def find(self, path, maxdepth=None, withdirs=False, detail=False, **kwargs):
+        if maxdepth is not None and maxdepth < 1:
+            raise ValueError("maxdepth must be at least 1")
+
+        # Remove the leading slash, as the zip file paths are always
+        # given without a leading slash
+        path = path.lstrip("/")
+        path_parts = list(filter(lambda s: bool(s), path.split("/")))
+
+        def _matching_starts(file_path):
+            file_parts = filter(lambda s: bool(s), file_path.split("/"))
+            return all(a == b for a, b in zip(path_parts, file_parts))
+
+        self._get_dirs()
+
+        result = {}
+        # To match posix find, if an exact file name is given, we should
+        # return only that file
+        if path in self.dir_cache and self.dir_cache[path]["type"] == "file":
+            result[path] = self.dir_cache[path]
+            return result if detail else [path]
+
+        for file_path, file_info in self.dir_cache.items():
+            if not (path == "" or _matching_starts(file_path)):
+                continue
+
+            if file_info["type"] == "directory":
+                if withdirs:
+                    if file_path not in result:
+                        result[file_path.strip("/")] = file_info
+                continue
+
+            if file_path not in result:
+                result[file_path] = file_info if detail else None
+
+        if maxdepth:
+            path_depth = path.count("/")
+            result = {
+                k: v for k, v in result.items() if k.count("/") - path_depth < maxdepth
+            }
+        return result if detail else sorted(result)