def parquet_dataset(metadata_path, schema=None, filesystem=None, format=None,
partitioning=None, partition_base_dir=None):
"""
Create a FileSystemDataset from a `_metadata` file created via
`pyarrrow.parquet.write_metadata`.
Parameters
----------
metadata_path : path,
Path pointing to a single file parquet metadata file
schema : Schema, optional
Optionally provide the Schema for the Dataset, in which case it will
not be inferred from the source.
filesystem : FileSystem or URI string, default None
If a single path is given as source and filesystem is None, then the
filesystem will be inferred from the path.
If an URI string is passed, then a filesystem object is constructed
using the URI's optional path component as a directory prefix. See the
examples below.
Note that the URIs on Windows must follow 'file:///C:...' or
'file:/C:...' patterns.
format : ParquetFileFormat
An instance of a ParquetFileFormat if special options needs to be
passed.
partitioning : Partitioning, PartitioningFactory, str, list of str
The partitioning scheme specified with the ``partitioning()``
function. A flavor string can be used as shortcut, and with a list of
field names a DirectionaryPartitioning will be inferred.
partition_base_dir : str, optional
For the purposes of applying the partitioning, paths will be
stripped of the partition_base_dir. Files not matching the
partition_base_dir prefix will be skipped for partitioning discovery.
The ignored files will still be part of the Dataset, but will not
have partition information.
Returns
-------
FileSystemDataset
"""
from pyarrow.fs import LocalFileSystem, _ensure_filesystem
if format is None:
format = ParquetFileFormat()
elif not isinstance(format, ParquetFileFormat):
raise ValueError("format argument must be a ParquetFileFormat")
if filesystem is None:
filesystem = LocalFileSystem()
else:
filesystem = _ensure_filesystem(filesystem)
metadata_path = filesystem.normalize_path(_stringify_path(metadata_path))
options = ParquetFactoryOptions(
partition_base_dir=partition_base_dir,
partitioning=_ensure_partitioning(partitioning)
)
factory = ParquetDatasetFactory(
metadata_path, filesystem, format, options=options)
return factory.finish(schema)
def _ensure_multiple_sources(paths, filesystem=None):
"""
Treat a list of paths as files belonging to a single file system
If the file system is local then also validates that all paths
are referencing existing *files* otherwise any non-file paths will be
silently skipped (for example on a remote filesystem).
Parameters
----------
paths : list of path-like
Note that URIs are not allowed.
filesystem : FileSystem or str, optional
If an URI is passed, then its path component will act as a prefix for
the file paths.
Returns
-------
(FileSystem, list of str)
File system object and a list of normalized paths.
Raises
------
TypeError
If the passed filesystem has wrong type.
IOError
If the file system is local and a referenced path is not available or
not a file.
"""
from pyarrow.fs import (LocalFileSystem, SubTreeFileSystem,
_MockFileSystem, FileType, _ensure_filesystem)
if filesystem is None:
# fall back to local file system as the default
filesystem = LocalFileSystem()
else:
# construct a filesystem if it is a valid URI
filesystem = _ensure_filesystem(filesystem)
is_local = (isinstance(filesystem, (LocalFileSystem, _MockFileSystem))
or (isinstance(filesystem, SubTreeFileSystem)
and isinstance(filesystem.base_fs, LocalFileSystem)))
# allow normalizing irregular paths such as Windows local paths
paths = [filesystem.normalize_path(_stringify_path(p)) for p in paths]
# validate that all of the paths are pointing to existing *files*
# possible improvement is to group the file_infos by type and raise for
# multiple paths per error category
if is_local:
for info in filesystem.get_file_info(paths):
file_type = info.type
if file_type == FileType.File:
continue
elif file_type == FileType.NotFound:
raise FileNotFoundError(info.path)
elif file_type == FileType.Directory:
raise IsADirectoryError(
'Path {} points to a directory, but only file paths are '
'supported. To construct a nested or union dataset pass '
'a list of dataset objects instead.'.format(info.path))
else:
raise IOError(
'Path {} exists but its type is unknown (could be a '
'special file such as a Unix socket or character device, '
'or Windows NUL / CON / ...)'.format(info.path))
return filesystem, paths
def _ensure_single_source(path, filesystem=None):
"""
Treat path as either a recursively traversable directory or a single file.
Parameters
----------
path : path-like
filesystem : FileSystem or str, optional
If an URI is passed, then its path component will act as a prefix for
the file paths.
Returns
-------
(FileSystem, list of str or fs.Selector)
File system object and either a single item list pointing to a file or
an fs.Selector object pointing to a directory.
Raises
------
TypeError
If the passed filesystem has wrong type.
FileNotFoundError
If the referenced file or directory doesn't exist.
"""
from pyarrow.fs import FileSystem, LocalFileSystem, FileType, FileSelector
path = _stringify_path(path)
# if filesystem is not given try to automatically determine one
# first check if the file exists as a local (relative) file path
# if not then try to parse the path as an URI
file_info = None
if filesystem is None:
filesystem = LocalFileSystem()
try:
file_info = filesystem.get_file_info([path])[0]
except OSError:
file_info = None
exists_locally = False
else:
exists_locally = (file_info.type != FileType.NotFound)
# if the file or directory doesn't exists locally, then assume that
# the path is an URI describing the file system as well
if not exists_locally:
try:
filesystem, path = FileSystem.from_uri(path)
except ValueError as e:
# ARROW-8213: neither an URI nor a locally existing path,
# so assume that local path was given and propagate a nicer
# file not found error instead of a more confusing scheme
# parsing error
if "empty scheme" not in str(e):
raise
else:
# unset file_info to query it again from the new filesystem
file_info = None
# construct a filesystem if it is a valid URI
filesystem, _ = _ensure_fs(filesystem)
# ensure that the path is normalized before passing to dataset discovery
path = filesystem.normalize_path(path)
# retrieve the file descriptor
if file_info is None:
file_info = filesystem.get_file_info([path])[0]
# depending on the path type either return with a recursive
# directory selector or as a list containing a single file
if file_info.type == FileType.Directory:
paths_or_selector = FileSelector(path, recursive=True)
elif file_info.type == FileType.File:
paths_or_selector = [path]
else:
raise FileNotFoundError(path)
return filesystem, paths_or_selector