Skip to main content

Minimalistic virtual filesystem adapters for Python

Project description

os: windows mac linux python: 3.10+ python style: google imports: isort code style: black code style: pycodestyle doc style: pydocstyle static typing: mypy linting: pylint testing: pytest security: bandit license: MIT

EZFS

EZFS (Easy Filesystem, or Everythingz a Filesystem) is an optimized, minimal dependency (down to 0), virtual filesystem adapter library for Python. EZFS provides access to the most common operations for "files" and "filesystems", while abstracting storage away from the user and developer, to simplify both UX and DX. All you need to know is how to access files in Python, and EZFS will take care of the rest, regardless of your selected backend.

In order to provide a streamlined UX and DX, EZFS leverages existing native Python designs and patterns, such as open/read/write operations, and applies them all backend storage types. The learning curve is minimal: if you know how to read/write a local file in Python, you also know how to read/write to any location in EZFS.

Reading and writing is supported for both text and binary files across local, remote, and memory "filesystems". Additional compression types and storage types can be supported by extending the primary File, Filesystem, and Compressor adapters. EZFS can also be faster than "native" open/read/write operations in some scenarios, due to having a specialized focus. Refer to the Compatibility guide, and Why EZFS?, for more information.

Table Of Contents

Compatibility

  • Supports Python 3.10+
  • Supports multiple storage types
    • Local filesystem
    • Temporary in-memory storage
    • sqlite3, local or in-memory (when built with Python)
    • S3 (when installed separately)
    • Any storage by extending File and Filesystem
  • Supports multiple compression types
    • bz2, gzip, lzma (when built with Python)
    • blosc, brotli, lz4, snappy, and zstd (when installed separately)
    • Any compression by extending Compressor or Transform
  • Custom transformations, such as encryption/decryption, Base64 encoding/decoding, obfuscation, etc.
    • Any data transformation by extending Transform

Getting Started

Installation

Install EZFS via pip:

pip install ezfs

Or via git clone:

git clone <path to fork>
cd ezfs
pip install .

Or build and install from wheel:

# Build locally.
git clone <path to fork>
cd ezfs
make wheel

# Push dist/ezfs*.tar.gz to environment where it will be installed.
pip install dist/ezfs*.tar.gz

Or via copy and paste (only a single file is required):

# Copy:
cp ezfs.py <target project directory>

How Tos

EZFS filesystems and file objects are designed to work nearly identical to native open() file handles. Basic read and write operations can be directly swapped out after creating a filesystem adapter, and calling open() against the filesystem instead of Python built-ins, or 3rd party compression libraries. Here are a few examples of how to use the more advanced features, such as compression and remote storage. Refer to the supported operations table in Why EZFS? for information on additional features.

Read or write a file

import ezfs

# No default compression/decompression:
fs = ezfs.LocalFilesystem('/tmp')

# With default compression/decompression for all files:
fs = ezfs.LocalFilesystem('/tmp', compression='gzip')

# Use default compression from filesystem during write:
with fs.open('test.txt.gz', 'w+') as out_file:
    out_file.write('test message')

# Manually specify compression during write:
with fs.open('test.txt.gz', 'w+', compression='gzip') as out_file:
    out_file.write('test message')

# Use default decompression from filesystem during read:
with fs.open('test.txt.gz') as in_file:
    print(in_file.read())

# Manually specify decompression during read:
with fs.open('test.txt.gz', compression='gzip') as in_file:
    print(in_file.read())

Swap between filesystem types (local file to local db)

import ezfs

# Only a single change is needed, such as from a local folder:
fs = ezfs.LocalFilesystem('/tmp')
# To a local database file:
fs = ezfs.SQLiteFilesystem('/tmp/tmp.db')

# No change is needed to open/read/write operations:
with fs.open('test.txt.gz', 'w+', compression='gzip') as out_file:
    out_file.write('test message')

Access a file (object) in an S3 bucket, and use compression

import ezfs

# To use advanced compression types, they must be installed separately.
fs = ezfs.S3BotoFilesystem(
    'my-bucket-1234',
    access_key_id='ABC123',
    secret_access_key='abcdefg1234567',
    compression='zstd',
)
with fs.open('test.txt.zst', 'w+') as out_file:
    out_file.write('test message')
with fs.open('test.txt.zst') as in_file:
    print(in_file.read())

Transform file data, such as Base64

import base64
import ezfs

b64_transform = ezfs.Transform(
    apply=lambda data: base64.b64encode(data),  # Used on write.
    remove=lambda data: base64.b64decode(data),  # Used on read.
)

# Transforms can be applied at the Filesystem level, or File level,
# similar to compression, with "transform=...":
fs = ezfs.LocalFilesystem('/tmp', transform=b64_transform)
with fs.open('test.txt', 'w+') as out_file:
    out_file.write('test message')
with fs.open('test.txt', transform=b64_transform) as in_file:
    print(in_file.read())

# Transforms can be combined to create complex transformations:
transform = ezfs.Transform.chain(
    b64_transform,
    ...
)

Why EZFS?

To simplify simple use cases.

EZFS is a very lightweight library (one file!), used to optimize "simple" use cases, or provide a starting point for more complex use cases. What make a use case "simple? Reliance on core file/filesystem functionality, such as create, read, write, and delete operations. What makes a use case "complex"? Reliance on complex file/filesystem features, such as permissions, streaming, and seeking. The former benefit from EZFS out-of-the-box, while the latter requires developers to extend the functionality further if they need this type of support.

While there are other libraries that can help accomplish file/filesystem-like use cases depending on the backend, such as s3fs for S3, they may be more than needed or wanted. For example, perhaps you have predictable logic to store/read files, and don't need to browse the filesystem tree. Perhaps you want to leverage a custom service to act as storage interchangeably with local files, without installing extra dependencies from other solutions. EZFS adapters can help with that. If you need full metadata support like filesystem tree browsing, or file permissions, EZFS cannot help with that (natively), and recommends using a more feature rich solution, or extending the adapters to fit your needs.

What does EZFS provide? What does EZFS not provide?

EZFS provides a shared, optimized, interface to read and write files to various backend locations, with or without compression. The backend for the storage can often be changed with a single line, without changing the rest of the code.

EZFS does not provide a complex feature set for advanced use cases, such as managing permissions or other metadata on filesystems. EZFS also does not provide streaming interfaces for processing larger than memory files in "chunks". The following is a list of common file/filesystem operations, whether they are supported out-of-the-box, whether they are supported with advanced installs (extras), and whether they are optimized/simplified by EZFS.

Operations OOB Extras Optimized Simplified
open() - ✅ ²
read() -
write() -
close() - ✅ ²
exists() - ✅ ³
isfile() - ✅ ³
remove() - ✅ ³
rename() - ✅ ³
Text files - ✅ ² ³
Binary files - ✅ ² ³
Local file storage -
Memory file storage -
S3 file storage
SQLite file storage ✅ ¹ -
Custom data transform -
bz2 compression ✅ ¹ -
gzip compression ✅ ¹ -
lzma compression ✅ ¹ -
blosc compression
brotli compression
lz4 compression
snappy compression
zstd compression
isdir() - -
listdir() - -
mkdir() - -
rmdir() - -
Other "os" calls - -
Other "os.path" calls - -
File permissions - -
File streaming - -
File seeking - -

¹ Depends on how Python was built
² Depends on compression module used
³ Depends on the backend used

Dependency Simplicity Example

Here is an example of using a library such as s3fs vs ezfs for basic read and write to S3, and its effect on required dependencies in a project. A basic boto3 install (only requirement for ezfs support) will add the following to the environment:

  • boto3
  • botocore
  • jmespath
  • python-dateutil
  • s3transfer
  • six
  • urllib3

An s3fs install will add the following in addition to the core boto3 requirements:

  • aiobotocore
  • aiohttp
  • aioitertools
  • aiosignal
  • async-timeout
  • attrs
  • idna
  • frozenlist
  • fsspec
  • multidict
  • s3fs
  • wrapt
  • yarl

Perhaps you already have all these requirements. Great! Then S3FS may be a better fit. Perhaps you don't have these, and want to reduce requirements that may add maintenance overhead to resolve security vulnerabilities. Great! EZFS may be a better fit. Still not sure? Continue reading for a performance example.

Optimized Remote Filesystem Example

Here is a basic performance example, using S3 + pandas to store DataFrames. EZFS can optimize the S3 client used to reduce networking overhead, leading to improved performance. The optimization benefit is greater with small files, but even larger files benefit, and the simplicity to use stays the same.

  • Small file: 100K
  • Large file: 1M
  • 100 iterations per test
  • s3fs 2024.2.0
  • ezfs 1.0.0
Scenario Write Read
pandas s3fs small raw 25.0 sec 15.6 sec
pandas ezfs small raw 16.9 sec 8.3 sec
pandas s3fs large raw 47.3 sec 17.9 sec
pandas ezfs large raw 28.7 sec 11.2 sec
pandas s3fs small zstd 20.2 sec 12.0 sec
pandas ezfs small zstd 12.6 sec 6.8 sec
pandas s3fs large zstd 37.4 sec 18.1 sec
pandas ezfs large zstd 21.5 sec 11.1 sec

Additional comparisons can be found in Benchmarks

Contributing

EZFS is not currently accepting new features. Minor features may be added to improve the native use cases, but outside minor changes it will only receive bug fixes and dependency updates. This decision is to ensure EZFS remains focused on its primary goal: stay simple and efficient, by focusing on simple use cases. Feel free to import, fork, copy, etc., to other projects to expand the scope of its ecosystem. Refer to the Contributing Guide for information on how to contribute fixes to this project.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

ezfs-1.1.1.tar.gz (20.4 kB view details)

Uploaded Source

Built Distribution

ezfs-1.1.1-py3-none-any.whl (15.1 kB view details)

Uploaded Python 3

File details

Details for the file ezfs-1.1.1.tar.gz.

File metadata

  • Download URL: ezfs-1.1.1.tar.gz
  • Upload date:
  • Size: 20.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.0.0 CPython/3.10.11

File hashes

Hashes for ezfs-1.1.1.tar.gz
Algorithm Hash digest
SHA256 3ee3939dcf744c84b29bfecf754916e5535f4d2b249df9a2a443340eac25fca5
MD5 eb2692e65ca1431a075377e134ad345c
BLAKE2b-256 90ab85abb9c10b0bb759e426f552492444c932b7da9b658d40dfe8546d6d8436

See more details on using hashes here.

File details

Details for the file ezfs-1.1.1-py3-none-any.whl.

File metadata

  • Download URL: ezfs-1.1.1-py3-none-any.whl
  • Upload date:
  • Size: 15.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.0.0 CPython/3.10.11

File hashes

Hashes for ezfs-1.1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 fbbd37b84901dd695aff262de423195b8fedd56f452c01096a724ffbd57479b7
MD5 dbccea077e420380701c35852acbe20a
BLAKE2b-256 fb2e11c18210b6680cddc71da13a0616750c21017222a4628f004f6281f2b0f0

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page