Skip to main content

Python 2.7 hashing and iteration in Python 3+

Project description

py27hash: Python 2.7 hashing and iteration in Python 3+

About

This package is designed to support Python 2.7 hashing and iteration in Python 3+. While there are a number of Python 2/3 compatibility libraries most are designed to support the old API classes/methods. This package will ensure the exact same behavior for dictionaries and sets

Python 2 will reach end of life (EOL) on January 1st, 2020. This package helps ease the migration from Python 2 to 3 for applications that depended on the old hash/iteration order of sets/dicts. Even when setting PYTHONHASHSEED=0, the hash (and default iteration order) will still be different as the hashing algorithm changed in Python3. Python 3.6 changed the default iteration order to insertion order.

One target use case is with machine learning. Optimization of machine learning model hyperparameters can take a very long time and if a model was built under Python 2 and feature set/dicts used the default sort order, new parameters would need to be used. This library can be used to allow a full conversion to Python 3 while fixes are made to re-optimize large model parameters.

This package implements logic in cpython 2.7 C source, mainly the Objects/ folder in pure Python. Performance was not a goal of this package and it will perform worse than native collections. It should only be used when there is a clear use case to preserve Python 2.7 hashing/iteration to ease a transition to Python 3.

Installation

The easiest way to install is via pip and PyPI pip install py27hash

You can also use Git to clone the repository from GitHub and install it manually:

git clone https://github.com/davidmezzetti/py27hash.git
cd py27hash
pip install .

Python 2.7 and 3+ are supported.

How to use

You only need to replace object instantiation to use this package. There are multiple ways to do this, with the best way to do it on a case by case basis.

Replace each instantiation

The first example replaces a single dict and set with a 2.7 dict.

from py27hash.dict import Dict

# Replace {} with Dict()
d = Dict()

d["a"] = 1

# Python 2.7 sets can be used in the same manner.

from py27hash.set import Set

# Replace set() with Set()
d = Set()

d.add("a")

Override function via import

The same example above can be changed to override the dict and set import statement. This will globally change an entire file which could cause issues but it is an option. If a dict or set was created with the {} syntax, it still needs to be changed to dict()/set().

from py27hash.dict import Dict as dict

d = dict()

d["a"] = 1

# Python 2.7 sets can be used in the same manner.
from py27hash.set import Set as set

d = set()

d.add("a")

Using other methods

The hashing and key iteration methods can be directly accessed via the hash and key packages as follows.

from py27hash.hash import hash27

print(hash27("test1234"))

# As with the example above you could even override the hash function for a particular file.

from p27hash.hash import hash27 as hash

print(hash("test1234"))

Both Dict and Set are backed by the keys class. As new values as added/modified, a Keys instance tracks each value to store the order via Python 2.7 hashing. This class can also be used directly.

from p27hash.key import Keys

keys = Keys()
keys.add("1")
keys.add("2")

Development

If an issue is found in this library, it can be cloned and changed.

git clone https://github.com/davidmezzetti/py27hash.git
cd py27hash

After changes are made to the source, unit tests should also be added and run via: scripts/test.sh

During development slower tests can be skipped via: scripts/test.sh skipslow

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

py27hash-1.0.0.tar.gz (7.5 kB view hashes)

Uploaded Source

Built Distribution

py27hash-1.0.0-py3-none-any.whl (9.5 kB view hashes)

Uploaded Python 3

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