Skip to content

A lightweight, high-performance Python library for parsing jsonl files.

License

Notifications You must be signed in to change notification settings

umarbutler/orjsonl

Repository files navigation

orjsonl

orjsonl is a lightweight, high-performance Python library for parsing jsonl files. It supports a wide variety of compression formats, including gzip, bzip2, xz and Zstandard. It is powered by orjson, the fastest and most accurate json serializer for Python.

Installation

orjsonl may be installed with pip:

pip install orjsonl

To read or write Zstandard files, install either zstd or the zstandard Python package.

Usage

The code snippet below demonstrates how jsonl files can be saved, loaded, streamed, appended and extended with orjsonl:

>>> import orjsonl
>>> # Create an iterable of Python objects.
>>> data = [
    'hello world',
    ['fizz', 'buzz'],
]
>>> # Save the iterable to a jsonl file.
>>> orjsonl.save('test.jsonl', data)
>>> # Append a Python object to the jsonl file.
>>> orjsonl.append('test.jsonl', {42 : 3.14})
>>> # Extend the jsonl file with an iterable of Python objects.
>>> orjsonl.extend('test.jsonl', [True, False])
>>> # Load the jsonl file.
>>> orjsonl.load('test.jsonl')
['hello world', ['fizz', 'buzz'], {42 : 3.14}, True, False]
>>> # Stream the jsonl file.
>>> list(orjsonl.stream('test.jsonl'))
['hello world', ['fizz', 'buzz'], {42 : 3.14}, True, False]

orjsonl can also be used to process jsonl files compressed with gzip, bzip2, xz and Zstandard:

>>> orjsonl.save('test.jsonl.gz', data)
>>> orjsonl.append('test.jsonl.gz', {42 : 3.14})
>>> orjsonl.extend('test.jsonl.gz', [True, False])
>>> orjsonl.load('test.jsonl.gz')
['hello world', ['fizz', 'buzz'], {42 : 3.14}, True, False]
>>> [obj for obj in orjsonl.stream('test.jsonl.gz')]
['hello world', ['fizz', 'buzz'], {42 : 3.14}, True, False]

Load

def load(
    path: str | bytes | int | os.PathLike,
    decompression_threads: Optional[int] = None,
    compression_format: Optional[str] = None
) -> list[dict | list | int | float | str | bool | None]

load() deserializes a compressed or uncompressed UTF-8-encoded jsonl file to a list of Python objects.

path is a path-like object giving the pathname (absolute or relative to the current working directory) of the compressed or uncompressed UTF-8-encoded jsonl file to be deserialized.

decompression_threads is an optional integer passed to xopen.xopen() as the threads argument that specifies the number of threads that should be used for decompression.

compression_format is an optional string passed to xopen.xopen() as the format argument that overrides the autodetection of the file’s compression format based on its extension or content. Possible values are ‘gz’, ‘xz’, ‘bz2’ and ‘zst’.

This function returns a list object comprised of deserialized dict, list, int, float, str, bool or None objects.

Stream

def stream(
    path: str | bytes | int | os.PathLike,
    decompression_threads: Optional[int] = None,
    compression_format: Optional[str] = None
) -> Generator[dict | list | int | float | str | bool | None, None, None]

stream() creates a generator that deserializes a compressed or uncompressed UTF-8-encoded jsonl file to Python objects.

path is a path-like object giving the pathname (absolute or relative to the current working directory) of the compressed or uncompressed UTF-8-encoded jsonl file to be deserialized by the generator.

decompression_threads is an optional integer passed to xopen.xopen() as the threads argument that specifies the number of threads that should be used for decompression.

compression_format is an optional string passed to xopen.xopen() as the format argument that overrides the autodetection of the file’s compression format based on its extension or content. Possible values are ‘gz’, ‘xz’, ‘bz2’ and ‘zst’.

This function returns a generator that deserializes the file to dict, list, int, float, str, bool or None objects.

Save

def save(
    path: str | bytes | int | os.PathLike,
    data: Iterable,
    default: Optional[Callable] = None,
    option: int = 0,
    compression_level: Optional[int] = None,
    compression_threads: Optional[int] = None,
    compression_format: Optional[str] = None
) -> None

save() serializes an iterable of Python objects to a compressed or uncompressed UTF-8-encoded jsonl file.

path is a path-like object giving the pathname (absolute or relative to the current working directory) of the compressed or uncompressed UTF-8-encoded jsonl file to be saved.

data is an iterable of Python objects to be serialized to the file.

default is an optional callable passed to orjson.dumps() as the default argument that serializes subclasses or arbitrary types to supported types.

option is an optional integer passed to orjson.dumps() as the option argument that modifies how data is serialized.

compression_level is an optional integer passed to xopen.xopen() as the compresslevel argument that determines the compression level for writing to gzip, xz and Zstandard files.

compression_threads is an optional integer passed to xopen.xopen() as the threads argument that specifies the number of threads that should be used for compression.

compression_format is an optional string passed to xopen.xopen() as the format argument that overrides the autodetection of the file’s compression format based on its extension. Possible values are ‘gz’, ‘xz’, ‘bz2’ and ‘zst’.

Append

def append(
    path: str | bytes | int | os.PathLike,
    data: Any,
    newline: bool = True,
    default: Optional[Callable] = None,
    option: int = 0,
    compression_level: Optional[int] = None,
    compression_threads: Optional[int] = None,
    compression_format: Optional[str] = None
) -> None

append() serializes and appends a Python object to a compressed or uncompressed UTF-8-encoded jsonl file.

path is a path-like object giving the pathname (absolute or relative to the current working directory) of the compressed or uncompressed UTF-8-encoded jsonl file to be appended.

data is a Python object to be serialized and appended to the file.

newline is an optional Boolean flag that, if set to False, indicates that the file does not end with a newline and should, therefore, have one added before data is appended.

default is an optional callable passed to orjson.dumps() as the default argument that serializes subclasses or arbitrary types to supported types.

option is an optional integer passed to orjson.dumps() as the option argument that modifies how data is serialized.

compression_level is an optional integer passed to xopen.xopen() as the compresslevel argument that determines the compression level for writing to gzip, xz and Zstandard files.

compression_threads is an optional integer passed to xopen.xopen() as the threads argument that specifies the number of threads that should be used for compression.

compression_format is an optional string passed to xopen.xopen() as the format argument that overrides the autodetection of the file’s compression format based on its extension or content. Possible values are ‘gz’, ‘xz’, ‘bz2’ and ‘zst’.

Extend

def extend(
    path: str | bytes | int | os.PathLike,
    data: Iterable,
    newline: bool = True,
    default: Optional[Callable] = None,
    option: int = 0,
    compression_level: Optional[int] = None,
    compression_threads: Optional[int] = None,
    compression_format: Optional[str] = None
) -> None

extend() serializes and appends an iterable of Python objects to a compressed or uncompressed UTF-8-encoded jsonl file.

path is a path-like object giving the pathname (absolute or relative to the current working directory) of the compressed or uncompressed UTF-8-encoded jsonl file to be extended.

data is an iterable of Python objects to be serialized and appended to the file.

newline is an optional Boolean flag that, if set to False, indicates that the file does not end with a newline and should, therefore, have one added before data is extended.

default is an optional callable passed to orjson.dumps() as the default argument that serializes subclasses or arbitrary types to supported types.

option is an optional integer passed to orjson.dumps() as the option argument that modifies how data is serialized.

compression_level is an optional integer passed to xopen.xopen() as the compresslevel argument that determines the compression level for writing to gzip, xz and Zstandard files.

compression_threads is an optional integer passed to xopen.xopen() as the threads argument that specifies the number of threads that should be used for compression.

compression_format is an optional string passed to xopen.xopen() as the format argument that overrides the autodetection of the file’s compression format based on its extension or content. Possible values are ‘gz’, ‘xz’, ‘bz2’ and ‘zst’.

License

This library is licensed under the MIT License.