Srsly: High-Performance Python Serialization Utilities

2.5.3 · active · verified Sun Apr 05

Srsly is a Python library providing modern, high-performance serialization utilities. It bundles optimized versions of popular serialization libraries (like ujson, msgpack, cloudpickle, and ruamel.yaml) into a single, standalone package, offering a consistent high-level API for handling JSON, JSONL, MessagePack, Pickle, and YAML data across different platforms and Python versions. It is actively maintained with frequent updates.

Warnings

breaking Srsly v2.x (and later) dropped support for Python 2.x. Users migrating from v1.x must ensure their environment is Python 3.6 or newer.
Fix: Upgrade to Python 3.6+ to use srsly v2.x. For Python 2.7, use srsly v1.x (which is no longer actively maintained).
gotcha Versions prior to v1.0.7 and early v2.x contained security vulnerabilities (CVE-2022-31116, CVE-2021-4595) related to incorrect handling of surrogate pair characters and buffer overflows in the vendored `ujson` component.
Fix: Upgrade to srsly v1.0.7 (if on 1.x) or v2.4.7 (or newer) to include the patches for these vulnerabilities.
gotcha Srsly v2.4.7 introduced a restriction to Cython 0.29.x due to incompatibilities with Cython 3. This primarily affects users who compile srsly from source or have Cython 3 installed in their environment.
Fix: Ensure that if you are building srsly from source, your Cython version is 0.29.x. If using Cython 3, you might encounter build issues until srsly officially supports it. Binary wheels are not affected by this specific build-time dependency.

Install

pip install srsly Install with pip

Imports

json_loads

import srsly
data = srsly.json_loads('{"key": "value"}')

read_json

import srsly
data = srsly.read_json('path/to/file.json')

write_json

import srsly
srsly.write_json('path/to/file.json', {'key': 'value'})

msgpack_dumps

import srsly
msg_bytes = srsly.msgpack_dumps({'key': 'value'})

Quickstart

This quickstart demonstrates how to use `srsly` to write and read data in JSON and JSONL formats. It covers basic file operations using `srsly.write_json`, `srsly.read_json`, `srsly.write_jsonl`, and `srsly.read_jsonl`.

import srsly
import os

# Example with JSON
file_path = 'example.json'
data = {"name": "Alice", "age": 30, "isStudent": False}

# Write JSON to a file
srsly.write_json(file_path, data)
print(f"Data written to {file_path}")

# Read JSON from a file
read_data = srsly.read_json(file_path)
print(f"Data read from {file_path}: {read_data}")

# Clean up
os.remove(file_path)

# Example with JSONL (newline-delimited JSON)
jsonl_path = 'example.jsonl'
lines = [{"id": 1, "text": "First line"}, {"id": 2, "text": "Second line"}]

# Write JSONL to a file
srsly.write_jsonl(jsonl_path, lines)
print(f"Data written to {jsonl_path}")

# Read JSONL from a file
read_lines = list(srsly.read_jsonl(jsonl_path))
print(f"Data read from {jsonl_path}: {read_lines}")

# Clean up
os.remove(jsonl_path)

view raw JSON →