Fast MUTF-8 encoder & decoder
Project Links
Meta
Author: Tyler Kennedy
Requires Python: >=3.9
Classifiers
Development Status
- 5 - Production/Stable
Intended Audience
- Developers
Operating System
- OS Independent
Programming Language
- C
- Python :: 3
- Python :: 3.9
- Python :: 3.10
- Python :: 3.11
- Python :: 3.12
- Python :: 3.13
- Python :: 3.14
- Python :: Implementation :: CPython
Topic
- Software Development :: Libraries :: Python Modules
mutf-8
This package contains simple pure-python as well as C encoders and decoders for the MUTF-8 character encoding.
These days, you'll most likely encounter MUTF-8 when working on files or
protocols related to the JVM. Strings in a Java .class file are encoded using
MUTF-8, strings passed by the JNI, as well as strings exported by the object
serializer.
This library was extracted from Lawu, a Python library for working with JVM class files.
Installation
Prebuilt binary packages are available for most platforms and Python versions. Install the package from PyPi:
pip install mutf8
Usage
Encoding and decoding is simple:
from mutf8 import encode_modified_utf8, decode_modified_utf8
unicode = decode_modified_utf8(byte_like_object)
bytes = encode_modified_utf8(unicode)
This module does not register itself globally as a codec, since importing should be side-effect-free.
Benchmarks
The C extension is dramatically faster than the pure-Python fallback — from ~10× on tiny strings (where per-call overhead dominates) to several hundred times on realistic payloads, and over 1000× when encoding large ASCII.
MUTF-8 Decoding
| Input | C | Python | Speedup |
|---|---|---|---|
| ASCII (16 B) | 81.5 ± 26 ns | 1.04 ± 0.18 µs | 13× |
| ASCII (1 KB) | 185 ± 38 ns | 65.7 ± 8.5 µs | 356× |
| ASCII (64 KB) | 5.58 ± 0.66 µs | 4.35 ± 0.23 ms | 780× |
| CJK (16 B) | 127 ± 15 ns | 1.33 ± 0.15 µs | 10× |
| CJK (1 KB) | 681 ± 117 ns | 99.3 ± 9.4 µs | 146× |
| CJK (64 KB) | 31.8 ± 3.3 µs | 6.28 ± 0.063 ms | 198× |
| Emoji (16 B) | 100 ± 21 ns | 882 ± 122 ns | 9× |
| Emoji (1 KB) | 538 ± 119 ns | 79.6 ± 6.9 µs | 148× |
| Emoji (64 KB) | 26.2 ± 4.1 µs | 5.32 ± 0.13 ms | 203× |
| Mixed (16 B) | 118 ± 27 ns | 2.17 ± 0.16 µs | 18× |
| Mixed (1 KB) | 860 ± 162 ns | 76.7 ± 9.2 µs | 89× |
| Mixed (64 KB) | 38 ± 5 µs | 5.3 ± 0.48 ms | 139× |
MUTF-8 Encoding
| Input | C | Python | Speedup |
|---|---|---|---|
| ASCII (16 B) | 70.5 ± 5.8 ns | 1.23 ± 0.15 µs | 18× |
| ASCII (1 KB) | 109 ± 16 ns | 49 ± 6.2 µs | 449× |
| ASCII (64 KB) | 1.55 ± 0.27 µs | 2.93 ± 0.19 ms | 1889× |
| CJK (16 B) | 81.4 ± 12 ns | 2.02 ± 0.26 µs | 25× |
| CJK (1 KB) | 624 ± 127 ns | 108 ± 16 µs | 173× |
| CJK (64 KB) | 32.5 ± 2.9 µs | 6.8 ± 0.24 ms | 209× |
| Emoji (16 B) | 73 ± 10 ns | 1.17 ± 0.12 µs | 16× |
| Emoji (1 KB) | 621 ± 76 ns | 67 ± 8.5 µs | 108× |
| Emoji (64 KB) | 33.3 ± 6 µs | 3.98 ± 0.12 ms | 119× |
| Mixed (16 B) | 108 ± 97 ns | 2.48 ± 0.35 µs | 23× |
| Mixed (1 KB) | 1.39 ± 0.29 µs | 63.3 ± 3.5 µs | 45× |
| Mixed (64 KB) | 81.2 ± 14 µs | 3.91 ± 0.093 ms | 48× |
C Extension
The C extension is optional. If a binary package is not available, or a C compiler is not present, the pure-python version will be used instead. If you want to ensure you're using the C version, import it directly:
from mutf8.cmutf8 import decode_modified_utf8
decode_modified_utf(b'\xED\xA1\x80\xED\xB0\x80')
1.1.0
Jul 10, 2026
1.0.6
Dec 29, 2021
1.0.5
Sep 11, 2021
1.0.4
Sep 10, 2021
1.0.3
Jan 20, 2021
1.0.2
Oct 12, 2020
1.0.1
Aug 26, 2020
1.0.0
Aug 26, 2020
Wheel compatibility matrix
Files in release
mutf8-1.1.0-cp310-cp310-macosx_10_9_x86_64.whl (12.9KiB)
mutf8-1.1.0-cp310-cp310-macosx_11_0_arm64.whl (13.2KiB)
mutf8-1.1.0-cp310-cp310-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl (22.9KiB)
mutf8-1.1.0-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl (23.9KiB)
mutf8-1.1.0-cp310-cp310-musllinux_1_2_aarch64.whl (23.3KiB)
mutf8-1.1.0-cp310-cp310-musllinux_1_2_x86_64.whl (22.8KiB)
mutf8-1.1.0-cp310-cp310-win32.whl (14.1KiB)
mutf8-1.1.0-cp310-cp310-win_amd64.whl (14.5KiB)
mutf8-1.1.0-cp311-cp311-macosx_11_0_arm64.whl (13.2KiB)
mutf8-1.1.0-cp311-cp311-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl (24.1KiB)
mutf8-1.1.0-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl (25.0KiB)
mutf8-1.1.0-cp311-cp311-musllinux_1_2_aarch64.whl (24.4KiB)
mutf8-1.1.0-cp311-cp311-musllinux_1_2_x86_64.whl (24.0KiB)
mutf8-1.1.0-cp311-cp311-win32.whl (14.2KiB)
mutf8-1.1.0-cp311-cp311-win_amd64.whl (14.5KiB)
mutf8-1.1.0-cp312-cp312-macosx_11_0_arm64.whl (13.2KiB)
mutf8-1.1.0-cp312-cp312-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl (24.3KiB)
mutf8-1.1.0-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl (25.2KiB)
mutf8-1.1.0-cp312-cp312-musllinux_1_2_aarch64.whl (24.6KiB)
mutf8-1.1.0-cp312-cp312-musllinux_1_2_x86_64.whl (24.2KiB)
mutf8-1.1.0-cp312-cp312-win32.whl (14.2KiB)
mutf8-1.1.0-cp312-cp312-win_amd64.whl (14.5KiB)
mutf8-1.1.0-cp313-cp313-macosx_11_0_arm64.whl (13.2KiB)
mutf8-1.1.0-cp313-cp313-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl (24.3KiB)
mutf8-1.1.0-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl (25.2KiB)
mutf8-1.1.0-cp313-cp313-musllinux_1_2_aarch64.whl (24.6KiB)
mutf8-1.1.0-cp313-cp313-musllinux_1_2_x86_64.whl (24.3KiB)
mutf8-1.1.0-cp313-cp313-win32.whl (14.2KiB)
mutf8-1.1.0-cp313-cp313-win_amd64.whl (14.5KiB)
mutf8-1.1.0-cp314-cp314-macosx_11_0_arm64.whl (13.2KiB)
mutf8-1.1.0-cp314-cp314-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl (24.4KiB)
mutf8-1.1.0-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl (25.3KiB)
mutf8-1.1.0-cp314-cp314-musllinux_1_2_aarch64.whl (24.7KiB)
mutf8-1.1.0-cp314-cp314-musllinux_1_2_x86_64.whl (24.3KiB)
mutf8-1.1.0-cp314-cp314-win32.whl (14.3KiB)
mutf8-1.1.0-cp314-cp314-win_amd64.whl (14.8KiB)
mutf8-1.1.0-cp314-cp314t-macosx_11_0_arm64.whl (13.3KiB)
mutf8-1.1.0-cp314-cp314t-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl (24.9KiB)
mutf8-1.1.0-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl (25.8KiB)
mutf8-1.1.0-cp314-cp314t-musllinux_1_2_aarch64.whl (25.2KiB)
mutf8-1.1.0-cp314-cp314t-musllinux_1_2_x86_64.whl (24.8KiB)
mutf8-1.1.0-cp314-cp314t-win32.whl (14.5KiB)
mutf8-1.1.0-cp314-cp314t-win_amd64.whl (14.9KiB)
mutf8-1.1.0-cp39-cp39-macosx_11_0_arm64.whl (13.2KiB)
mutf8-1.1.0-cp39-cp39-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl (22.7KiB)
mutf8-1.1.0-cp39-cp39-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl (23.7KiB)
mutf8-1.1.0-cp39-cp39-musllinux_1_2_aarch64.whl (23.2KiB)
mutf8-1.1.0-cp39-cp39-musllinux_1_2_x86_64.whl (22.6KiB)
mutf8-1.1.0-cp39-cp39-win32.whl (14.1KiB)
mutf8-1.1.0-cp39-cp39-win_amd64.whl (14.5KiB)
mutf8-1.1.0.tar.gz (11.5KiB)
