iotools-wrappers 0.1.0

Provides a unified syntax for reading and writing data to files

iotools

iotools provides a unified syntax for reading and writing data to files.

Example:

text = load_txt("filename.txt")
data = load_json("filename.json")
table = load_csv("filename.csv")
image = load_jpg("filename.jpg")
config = load_yaml("filename.yaml")

Functions

Reading

• data = load_*(filename): filename is a path to an existing file
• data = read_*(openfile): openfile is a file-like object
• data = decode_*(rawbytes): rawbytes is the file's bytes

Writing

• save_*(filename, data): filename is a path to an existing file
• write_*(openfile, data): openfile is a file-like object
• rawbytes = encode_*(data): returns bytes

Helping

• help_*(): prints simple examples and simple implementations on how to use this file format without iotools functions

Installation

pip install iotools-wrappers

Supported formats

• avi
• bytes: not really a format, but use this to load raw bytes
• csv
• gif
• gz: gzip
• jpg
• json
• mp4
• npy: numpy
• pickle
• png
• tar
• txt: text files
• xml
• yaml
• zip
• zst: zstandard

How it works

iotools is nothing but wrappers on other python librairies, for example load_json will call json.load internally. The backend used is specified in each function documentation.

iotools goal is to provide a memorable io syntax, on as many file formats as possible.

For more format-specific functions, or more complex io features, please use the library specific to that file format (help_*() may provide a good starting point)

Examples

Text

from iotools import load_txt, read_txt

# Load text in one line:
text = load_txt("filename.txt")

# Or if you prefer the 'with open()' syntax:
with open("filename.txt", "r") as f:
     text = read_txt(f)  # same as f.read()

Images

# Example where image colors are inversed
from iotools import load_image, save_image

image = load_image("input.jpg")
save_image("output.jpg", image[:, :, ::-1])

Archives

# Load a tar.gz archive
from iotools import load_gz, decode_tar

data = decode_tar(load_gz("archive.tar.gz"))

# If your archive is made of images, you can even do:
from iotools import load_gz, decode_tar, decode_image

data = {filename: decode_image(img) for filename, img in decode_tar(load_gz("archive.tar.gz")).items()}

GIFs

import numpy as np
from iotools import save_gif

gif = [np.ones((180, 320), dtype=np.uint8) * i * 10 for i in range(26)]
save_gif("filename.gif", gif, fps=6)

Additionnal features

File systems

iotools is compatible with fsspec: load_* and save_* functions have an fs kwarg in order to use different file systems.

Example:

# Load a tar.gz archive containing images
from iotools import load_image
from fsspec.implementations.tar import TarFileSystem

tarfs = TarFileSystem("archive.tar.gz")
data = {filename: load_image(filename, fs=tarfs) for filename in tarfs.glob("*")}

Makedirs

save_*(filename, data, fs=None, makedirs=False) have a kwarg makedirs that allows to automatically create filename's parent directory if it does not exist.

Format specific options

Since iotools functions are mostly wrappers of other python libraires, most of formats specific options can be used through iotools functions. For example you can do save_json("filename.json", data, indent=4).

Use your prefered library

Sometimes several python librairies can be used to load/save data to a format. To let you use your preferred library, iotools functions are declined in several forms with the keyword using. Here is an example with jpg images:

from iotools.jpgio import load_jpg_using_imageio, load_jpg_using_pil, load_jpg_using_cv2

image = load_jpg_using_imageio("filename.jpg")  # uses imageio library
image = load_jpg_using_pil("filename.jpg")  # uses PIL (Pillow) library
image = load_jpg_using_cv2("filename.jpg")  # uses cv2 (opencv-python) library

Note that the output image can be different based on which library is used internally

Generic formats

iotools supports 2 generic formats: images and videos. Thus there are functions like load_image or load_video to load any image or video format (formats not enumerated above were not tested). save_image or save_video also exists, and a kwarg format is there for you to choose which image/video format to save in, otherwise it will be guessed from the filename.

Other notes

Import syntax

Several import syntaxs can be used:

from iotools import load_csv
from iotools import csvio  # --> csvio.load_csv
from iotools.csvio import load_csv, load_csv_using_pandas

OS Support

iotools was only tested under Linux. Under Windows, it seems that there are still issues with csvs, images and videos, but the other formats should be fine.

Dependencies

Installing iotools does not install any format-specific backends, thus you may not be able to use a file format if the backend is not already installed. Please refer to pyproject.toml for backends supported versions.