def write_dataset(data,
base_dir,
basename_template=None,
format=None,
partitioning=None,
schema=None,
filesystem=None,
file_options=None,
use_threads=True,
use_async=False,
max_partitions=None,
file_visitor=None):
"""
Write a dataset to a given format and partitioning.
Parameters
----------
data : Dataset, Table/RecordBatch, RecordBatchReader, list of
Table/RecordBatch, or iterable of RecordBatch
The data to write. This can be a Dataset instance or
in-memory Arrow data. If an iterable is given, the schema must
also be given.
base_dir : str
The root directory where to write the dataset.
basename_template : str, optional
A template string used to generate basenames of written data files.
The token '{i}' will be replaced with an automatically incremented
integer. If not specified, it defaults to
"part-{i}." + format.default_extname
format : FileFormat or str
The format in which to write the dataset. Currently supported:
"parquet", "ipc"/"feather". If a FileSystemDataset is being written
and `format` is not specified, it defaults to the same format as the
specified FileSystemDataset. When writing a Table or RecordBatch, this
keyword is required.
partitioning : Partitioning, optional
The partitioning scheme specified with the ``partitioning()``
function.
schema : Schema, optional
filesystem : FileSystem, optional
file_options : FileWriteOptions, optional
FileFormat specific write options, created using the
``FileFormat.make_write_options()`` function.
use_threads : bool, default True
Write files in parallel. If enabled, then maximum parallelism will be
used determined by the number of available CPU cores.
use_async : bool, default False
If enabled, an async scanner will be used that should offer
better performance with high-latency/highly-parallel filesystems
(e.g. S3)
max_partitions : int, default 1024
Maximum number of partitions any batch may be written into.
file_visitor : Function
If set, this function will be called with a WrittenFile instance
for each file created during the call. This object will have both
a path attribute and a metadata attribute.
The path attribute will be a string containing the path to
the created file.
The metadata attribute will be the parquet metadata of the file.
This metadata will have the file path attribute set and can be used
to build a _metadata file. The metadata attribute will be None if
the format is not parquet.
Example visitor which simple collects the filenames created::
visited_paths = []
def file_visitor(written_file):
visited_paths.append(written_file.path)
"""
from pyarrow.fs import _resolve_filesystem_and_path
if isinstance(data, (list, tuple)):
schema = schema or data[0].schema
data = InMemoryDataset(data, schema=schema)
elif isinstance(data, (pa.RecordBatch, pa.Table)):
schema = schema or data.schema
data = InMemoryDataset(data, schema=schema)
elif isinstance(data, pa.ipc.RecordBatchReader) or _is_iterable(data):
data = Scanner.from_batches(data, schema=schema)
schema = None
elif not isinstance(data, (Dataset, Scanner)):
raise ValueError(
"Only Dataset, Scanner, Table/RecordBatch, RecordBatchReader, "
"a list of Tables/RecordBatches, or iterable of batches are "
"supported.")
if format is None and isinstance(data, FileSystemDataset):
format = data.format
else:
format = _ensure_format(format)
if file_options is None:
file_options = format.make_write_options()
if format != file_options.format:
raise TypeError("Supplied FileWriteOptions have format {}, "
"which doesn't match supplied FileFormat {}".format(
format, file_options))
if basename_template is None:
basename_template = "part-{i}." + format.default_extname
if max_partitions is None:
max_partitions = 1024
partitioning = _ensure_write_partitioning(partitioning)
filesystem, base_dir = _resolve_filesystem_and_path(base_dir, filesystem)
if isinstance(data, Dataset):
scanner = data.scanner(use_threads=use_threads, use_async=use_async)
else:
# scanner was passed directly by the user, in which case a schema
# cannot be passed
if schema is not None:
raise ValueError("Cannot specify a schema when writing a Scanner")
scanner = data
_filesystemdataset_write(scanner, base_dir, basename_template, filesystem,
partitioning, file_options, max_partitions,
file_visitor)
def dataset(source,
schema=None,
format=None,
filesystem=None,
partitioning=None,
partition_base_dir=None,
exclude_invalid_files=None,
ignore_prefixes=None):
"""
Open a dataset.
Datasets provides functionality to efficiently work with tabular,
potentially larger than memory and multi-file dataset.
- A unified interface for different sources, like Parquet and Feather
- Discovery of sources (crawling directories, handle directory-based
partitioned datasets, basic schema normalization)
- Optimized reading with predicate pushdown (filtering rows), projection
(selecting columns), parallel reading or fine-grained managing of tasks.
Note that this is the high-level API, to have more control over the dataset
construction use the low-level API classes (FileSystemDataset,
FilesystemDatasetFactory, etc.)
Parameters
----------
source : path, list of paths, dataset, list of datasets, (list of) batches\
or tables, iterable of batches, RecordBatchReader, or URI
Path pointing to a single file:
Open a FileSystemDataset from a single file.
Path pointing to a directory:
The directory gets discovered recursively according to a
partitioning scheme if given.
List of file paths:
Create a FileSystemDataset from explicitly given files. The files
must be located on the same filesystem given by the filesystem
parameter.
Note that in contrary of construction from a single file, passing
URIs as paths is not allowed.
List of datasets:
A nested UnionDataset gets constructed, it allows arbitrary
composition of other datasets.
Note that additional keyword arguments are not allowed.
(List of) batches or tables, iterable of batches, or RecordBatchReader:
Create an InMemoryDataset. If an iterable or empty list is given,
a schema must also be given. If an iterable or RecordBatchReader
is given, the resulting dataset can only be scanned once; further
attempts will raise an error.
schema : Schema, optional
Optionally provide the Schema for the Dataset, in which case it will
not be inferred from the source.
format : FileFormat or str
Currently "parquet" and "ipc"/"arrow"/"feather" are supported. For
Feather, only version 2 files are supported.
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.
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.
exclude_invalid_files : bool, optional (default True)
If True, invalid files will be excluded (file format specific check).
This will incur IO for each files in a serial and single threaded
fashion. Disabling this feature will skip the IO, but unsupported
files may be present in the Dataset (resulting in an error at scan
time).
ignore_prefixes : list, optional
Files matching any of these prefixes will be ignored by the
discovery process. This is matched to the basename of a path.
By default this is ['.', '_'].
Note that discovery happens only if a directory is passed as source.
Returns
-------
dataset : Dataset
Either a FileSystemDataset or a UnionDataset depending on the source
parameter.
Examples
--------
Opening a single file:
>>> dataset("path/to/file.parquet", format="parquet")
Opening a single file with an explicit schema:
>>> dataset("path/to/file.parquet", schema=myschema, format="parquet")
Opening a dataset for a single directory:
>>> dataset("path/to/nyc-taxi/", format="parquet")
>>> dataset("s3://mybucket/nyc-taxi/", format="parquet")
Opening a dataset from a list of relatives local paths:
>>> dataset([
... "part0/data.parquet",
... "part1/data.parquet",
... "part3/data.parquet",
... ], format='parquet')
With filesystem provided:
>>> paths = [
... 'part0/data.parquet',
... 'part1/data.parquet',
... 'part3/data.parquet',
... ]
>>> dataset(paths, filesystem='file:///directory/prefix, format='parquet')
Which is equivalent with:
>>> fs = SubTreeFileSystem("/directory/prefix", LocalFileSystem())
>>> dataset(paths, filesystem=fs, format='parquet')
With a remote filesystem URI:
>>> paths = [
... 'nested/directory/part0/data.parquet',
... 'nested/directory/part1/data.parquet',
... 'nested/directory/part3/data.parquet',
... ]
>>> dataset(paths, filesystem='s3://bucket/', format='parquet')
Similarly to the local example, the directory prefix may be included in the
filesystem URI:
>>> dataset(paths, filesystem='s3://bucket/nested/directory',
... format='parquet')
Construction of a nested dataset:
>>> dataset([
... dataset("s3://old-taxi-data", format="parquet"),
... dataset("local/path/to/data", format="ipc")
... ])
"""
# collect the keyword arguments for later reuse
kwargs = dict(schema=schema,
filesystem=filesystem,
partitioning=partitioning,
format=format,
partition_base_dir=partition_base_dir,
exclude_invalid_files=exclude_invalid_files,
selector_ignore_prefixes=ignore_prefixes)
if _is_path_like(source):
return _filesystem_dataset(source, **kwargs)
elif isinstance(source, (tuple, list)):
if all(_is_path_like(elem) for elem in source):
return _filesystem_dataset(source, **kwargs)
elif all(isinstance(elem, Dataset) for elem in source):
return _union_dataset(source, **kwargs)
elif all(
isinstance(elem, (pa.RecordBatch, pa.Table))
for elem in source):
return _in_memory_dataset(source, **kwargs)
else:
unique_types = set(type(elem).__name__ for elem in source)
type_names = ', '.join('{}'.format(t) for t in unique_types)
raise TypeError(
'Expected a list of path-like or dataset objects, or a list '
'of batches or tables. The given list contains the following '
'types: {}'.format(type_names))
elif isinstance(source,
(pa.RecordBatch, pa.ipc.RecordBatchReader, pa.Table)):
return _in_memory_dataset(source, **kwargs)
elif _is_iterable(source):
return _in_memory_dataset(source, **kwargs)
else:
raise TypeError(
'Expected a path-like, list of path-likes or a list of Datasets '
'instead of the given type: {}'.format(type(source).__name__))
def write_dataset(data,
base_dir,
basename_template=None,
format=None,
partitioning=None,
schema=None,
filesystem=None,
file_options=None,
use_threads=True,
max_partitions=None):
"""
Write a dataset to a given format and partitioning.
Parameters
----------
data : Dataset, Table/RecordBatch, RecordBatchReader, list of
Table/RecordBatch, or iterable of RecordBatch
The data to write. This can be a Dataset instance or
in-memory Arrow data. If an iterable is given, the schema must
also be given.
base_dir : str
The root directory where to write the dataset.
basename_template : str, optional
A template string used to generate basenames of written data files.
The token '{i}' will be replaced with an automatically incremented
integer. If not specified, it defaults to
"part-{i}." + format.default_extname
format : FileFormat or str
The format in which to write the dataset. Currently supported:
"parquet", "ipc"/"feather". If a FileSystemDataset is being written
and `format` is not specified, it defaults to the same format as the
specified FileSystemDataset. When writing a Table or RecordBatch, this
keyword is required.
partitioning : Partitioning, optional
The partitioning scheme specified with the ``partitioning()``
function.
schema : Schema, optional
filesystem : FileSystem, optional
file_options : FileWriteOptions, optional
FileFormat specific write options, created using the
``FileFormat.make_write_options()`` function.
use_threads : bool, default True
Write files in parallel. If enabled, then maximum parallelism will be
used determined by the number of available CPU cores.
max_partitions : int, default 1024
Maximum number of partitions any batch may be written into.
"""
from pyarrow.fs import _resolve_filesystem_and_path
if isinstance(data, Dataset):
schema = schema or data.schema
elif isinstance(data, (list, tuple)):
schema = schema or data[0].schema
data = InMemoryDataset(data, schema=schema)
elif isinstance(data, (pa.RecordBatch, pa.ipc.RecordBatchReader,
pa.Table)) or _is_iterable(data):
data = InMemoryDataset(data, schema=schema)
schema = schema or data.schema
else:
raise ValueError(
"Only Dataset, Table/RecordBatch, RecordBatchReader, a list "
"of Tables/RecordBatches, or iterable of batches are supported.")
if format is None and isinstance(data, FileSystemDataset):
format = data.format
else:
format = _ensure_format(format)
if file_options is None:
file_options = format.make_write_options()
if format != file_options.format:
raise TypeError("Supplied FileWriteOptions have format {}, "
"which doesn't match supplied FileFormat {}".format(
format, file_options))
if basename_template is None:
basename_template = "part-{i}." + format.default_extname
if max_partitions is None:
max_partitions = 1024
partitioning = _ensure_write_partitioning(partitioning)
filesystem, base_dir = _resolve_filesystem_and_path(base_dir, filesystem)
_filesystemdataset_write(data, base_dir, basename_template, schema,
filesystem, partitioning, file_options,
use_threads, max_partitions)
def write_dataset(data, base_dir, basename_template=None, format=None,
partitioning=None, partitioning_flavor=None, schema=None,
filesystem=None, file_options=None, use_threads=True,
max_partitions=None, max_open_files=None,
max_rows_per_file=None, min_rows_per_group=None,
max_rows_per_group=None, file_visitor=None,
existing_data_behavior='error', create_dir=True):
"""
Write a dataset to a given format and partitioning.
Parameters
----------
data : Dataset, Table/RecordBatch, RecordBatchReader, list of \
Table/RecordBatch, or iterable of RecordBatch
The data to write. This can be a Dataset instance or
in-memory Arrow data. If an iterable is given, the schema must
also be given.
base_dir : str
The root directory where to write the dataset.
basename_template : str, optional
A template string used to generate basenames of written data files.
The token '{i}' will be replaced with an automatically incremented
integer. If not specified, it defaults to
"part-{i}." + format.default_extname
format : FileFormat or str
The format in which to write the dataset. Currently supported:
"parquet", "ipc"/"arrow"/"feather", and "csv". If a FileSystemDataset
is being written and `format` is not specified, it defaults to the
same format as the specified FileSystemDataset. When writing a
Table or RecordBatch, this keyword is required.
partitioning : Partitioning or list[str], optional
The partitioning scheme specified with the ``partitioning()``
function or a list of field names. When providing a list of
field names, you can use ``partitioning_flavor`` to drive which
partitioning type should be used.
partitioning_flavor : str, optional
One of the partitioning flavors supported by
``pyarrow.dataset.partitioning``. If omitted will use the
default of ``partitioning()`` which is directory partitioning.
schema : Schema, optional
filesystem : FileSystem, optional
file_options : pyarrow.dataset.FileWriteOptions, optional
FileFormat specific write options, created using the
``FileFormat.make_write_options()`` function.
use_threads : bool, default True
Write files in parallel. If enabled, then maximum parallelism will be
used determined by the number of available CPU cores.
max_partitions : int, default 1024
Maximum number of partitions any batch may be written into.
max_open_files : int, default 1024
If greater than 0 then this will limit the maximum number of
files that can be left open. If an attempt is made to open
too many files then the least recently used file will be closed.
If this setting is set too low you may end up fragmenting your
data into many small files.
max_rows_per_file : int, default 0
Maximum number of rows per file. If greater than 0 then this will
limit how many rows are placed in any single file. Otherwise there
will be no limit and one file will be created in each output
directory unless files need to be closed to respect max_open_files
min_rows_per_group : int, default 0
Minimum number of rows per group. When the value is greater than 0,
the dataset writer will batch incoming data and only write the row
groups to the disk when sufficient rows have accumulated.
max_rows_per_group : int, default 1024 * 1024
Maximum number of rows per group. If the value is greater than 0,
then the dataset writer may split up large incoming batches into
multiple row groups. If this value is set, then min_rows_per_group
should also be set. Otherwise it could end up with very small row
groups.
file_visitor : function
If set, this function will be called with a WrittenFile instance
for each file created during the call. This object will have both
a path attribute and a metadata attribute.
The path attribute will be a string containing the path to
the created file.
The metadata attribute will be the parquet metadata of the file.
This metadata will have the file path attribute set and can be used
to build a _metadata file. The metadata attribute will be None if
the format is not parquet.
Example visitor which simple collects the filenames created::
visited_paths = []
def file_visitor(written_file):
visited_paths.append(written_file.path)
existing_data_behavior : 'error' | 'overwrite_or_ignore' | \
'delete_matching'
Controls how the dataset will handle data that already exists in
the destination. The default behavior ('error') is to raise an error
if any data exists in the destination.
'overwrite_or_ignore' will ignore any existing data and will
overwrite files with the same name as an output file. Other
existing files will be ignored. This behavior, in combination
with a unique basename_template for each write, will allow for
an append workflow.
'delete_matching' is useful when you are writing a partitioned
dataset. The first time each partition directory is encountered
the entire directory will be deleted. This allows you to overwrite
old partitions completely.
create_dir : bool, default True
If False, directories will not be created. This can be useful for
filesystems that do not require directories.
"""
from pyarrow.fs import _resolve_filesystem_and_path
if isinstance(data, (list, tuple)):
schema = schema or data[0].schema
data = InMemoryDataset(data, schema=schema)
elif isinstance(data, (pa.RecordBatch, pa.Table)):
schema = schema or data.schema
data = InMemoryDataset(data, schema=schema)
elif isinstance(data, pa.ipc.RecordBatchReader) or _is_iterable(data):
data = Scanner.from_batches(data, schema=schema)
schema = None
elif not isinstance(data, (Dataset, Scanner)):
raise ValueError(
"Only Dataset, Scanner, Table/RecordBatch, RecordBatchReader, "
"a list of Tables/RecordBatches, or iterable of batches are "
"supported."
)
if format is None and isinstance(data, FileSystemDataset):
format = data.format
else:
format = _ensure_format(format)
if file_options is None:
file_options = format.make_write_options()
if format != file_options.format:
raise TypeError("Supplied FileWriteOptions have format {}, "
"which doesn't match supplied FileFormat {}".format(
format, file_options))
if basename_template is None:
basename_template = "part-{i}." + format.default_extname
if max_partitions is None:
max_partitions = 1024
if max_open_files is None:
max_open_files = 1024
if max_rows_per_file is None:
max_rows_per_file = 0
if max_rows_per_group is None:
max_rows_per_group = 1 << 20
if min_rows_per_group is None:
min_rows_per_group = 0
# at this point data is a Scanner or a Dataset, anything else
# was converted to one of those two. So we can grab the schema
# to build the partitioning object from Dataset.
if isinstance(data, Scanner):
partitioning_schema = data.dataset_schema
else:
partitioning_schema = data.schema
partitioning = _ensure_write_partitioning(partitioning,
schema=partitioning_schema,
flavor=partitioning_flavor)
filesystem, base_dir = _resolve_filesystem_and_path(base_dir, filesystem)
if isinstance(data, Dataset):
scanner = data.scanner(use_threads=use_threads)
else:
# scanner was passed directly by the user, in which case a schema
# cannot be passed
if schema is not None:
raise ValueError("Cannot specify a schema when writing a Scanner")
scanner = data
_filesystemdataset_write(
scanner, base_dir, basename_template, filesystem, partitioning,
file_options, max_partitions, file_visitor, existing_data_behavior,
max_open_files, max_rows_per_file,
min_rows_per_group, max_rows_per_group, create_dir
)