# Copyright 2023 Iguazio
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import pathlib
from fsspec.implementations.dbfs import DatabricksFile, DatabricksFileSystem
from fsspec.registry import get_filesystem_class
import mlrun.errors
from .base import DataStore, FileStats, make_datastore_schema_sanitizer
[docs]class DatabricksFileBugFixed(DatabricksFile):
"""Overrides DatabricksFile to add the following fix: https://github.com/fsspec/filesystem_spec/pull/1278"""
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(end=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""
for chunk_start, chunk_end in self._to_sized_blocks(start, end):
return_buffer += self.fs._get_data(
path=self.path, start=chunk_start, end=chunk_end
)
return return_buffer
def _to_sized_blocks(self, start=0, end=100):
"""Helper function to split a range from 0 to total_length into blocksizes"""
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
[docs]class DatabricksFileSystemDisableCache(DatabricksFileSystem):
root_marker = "/"
protocol = "dbfs"
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 DatabricksFileBugFixed(
self, path, mode=mode, block_size=block_size, **kwargs
)
# _ls_from_cache is not working properly, so we disable it.
def _ls_from_cache(self, path):
pass
# dbfs objects will be represented with the following URL: dbfs://<path>
class DBFSStore(DataStore):
def __init__(self, parent, schema, name, endpoint="", secrets: dict = None):
super().__init__(parent, name, schema, endpoint, secrets=secrets)
@property
def filesystem(self):
"""return fsspec file system object, if supported"""
filesystem_class = get_filesystem_class(protocol=self.kind)
if not self._filesystem:
self._filesystem = make_datastore_schema_sanitizer(
cls=filesystem_class,
using_bucket=False,
**self.get_storage_options(),
)
return self._filesystem
def get_storage_options(self):
res = dict(
token=self._get_secret_or_env("DATABRICKS_TOKEN"),
instance=self._get_secret_or_env("DATABRICKS_HOST"),
)
return self._sanitize_storage_options(res)
def _verify_filesystem_and_key(self, key: str):
if not self.filesystem:
raise mlrun.errors.MLRunInvalidArgumentError(
"Performing actions on data-item without a valid filesystem"
)
if not key.startswith("/"):
raise mlrun.errors.MLRunInvalidArgumentError(
"Invalid key parameter - key must start with '/'"
)
def get(self, key: str, size=None, offset=0) -> bytes:
self._verify_filesystem_and_key(key)
if size is not None and size < 0:
raise mlrun.errors.MLRunInvalidArgumentError("size cannot be negative")
if offset is None:
raise mlrun.errors.MLRunInvalidArgumentError("offset cannot be None")
start = offset or None
end = offset + size if size else None
return self.filesystem.cat_file(key, start=start, end=end)
def put(self, key, data, append=False):
self._verify_filesystem_and_key(key)
if append:
raise mlrun.errors.MLRunInvalidArgumentError(
"Append mode not supported for Databricks file system"
)
# can not use append mode because it overrides data.
data, mode = self._prepare_put_data(data, append)
with self.filesystem.open(key, mode) as f:
f.write(data)
def upload(self, key: str, src_path: str):
self._verify_filesystem_and_key(key)
self.filesystem.put_file(src_path, key, overwrite=True)
def stat(self, key: str):
self._verify_filesystem_and_key(key)
file = self.filesystem.stat(key)
if file["type"] == "file":
size = file["size"]
elif file["type"] == "directory":
raise FileNotFoundError("Operation expects a file not a directory!")
return FileStats(size, None)
def listdir(self, key: str):
"""
Basic ls of file/dir - without recursion.
"""
self._verify_filesystem_and_key(key)
if self.filesystem.isfile(key):
return key
remote_path = f"{key}/*"
files = self.filesystem.glob(remote_path)
# Get only the files and directories under key path, without the key path itself.
# for example in a filesystem that has this path: /test_mlrun_dbfs_objects/test.txt
# listdir with the input /test_mlrun_dbfs_objects as a key will return ['test.txt'].
files = [pathlib.Path(file).name for file in files if "/" in file]
return files
def rm(self, path, recursive=False, maxdepth=None):
if maxdepth:
raise mlrun.errors.MLRunInvalidArgumentError(
"dbfs file system does not support maxdepth option in rm function"
)
self.filesystem.rm(path=path, recursive=recursive)
