Apertium + Python

Introduction

• The codebase is in development for the GSoC '19 project called Apertium API in Python
• The Apertium core modules are written in C++.
• This project makes the Apertium modules available in Python, which because of its simplicity is more appealing to users.

About the Exisiting Code Base

• The existing codebase has Subprocess and SWIG wrapper implementations of the higher level functions of Apertium and CG modules.

Installation

• Installation on Debian/Ubuntu and Windows is natively supported:

  pip install apertium

• For developers, pipenv can be used to install the development dependencies and enter a shell with them:

  pip install pipenv
  pipenv install --dev
  pipenv shell

• Apertium packages can be installed from Python interpreter as well.

  • Install apertium-all-dev by calling apertium.installer.install_apertium()
  • Install language packages with apertium.installer.install_module(language_name). For example apertium-eng can be installed by executing apertium.installer.install_module('eng')

Usage

• For multiple invocations Method 1 is more performant, as the dictionary needs to be loaded only once.

Analysis

Performing Morphological Analysis

Method 1: Create an Analyzer object and call its analyze method.

In [1]: import apertium
In [2]: a = apertium.Analyzer('en')
In [3]: a.analyze('cats')
Out[3]: [cats/cat<n><pl>, ./.<sent>]

Method 2: Calling analyze() directly.

In [1]: import apertium
In [2]: apertium.analyze('en', 'cats')
Out[2]: cats/cat<n><pl>

Generation

Performing Morphological Generation

Method 1: Create a Generator object and call its generate method.

In [1]: import apertium
In [2]: g = apertium.Generator('en')
In [3]: g.generate('^cat<n><pl>$')
Out[3]: 'cats'

Method 2: Calling generate() directly.

In [1]: import apertium
In [2]: apertium.generate('en', '^cat<n><pl>$')
Out[2]: 'cats'

Tagger

Method 1: Create a Tagger object and call its tag method.

In [1]: import apertium
In [2]: tagger = apertium.Tagger('eng')
In [3]: tagger.tag('cats')
Out[3]: [cats/cat<n><pl>]

Method 2: Calling tag() directly.

In [1]: import apertium
In [2]: apertium.tag('en', 'cats')
Out[2]: [cats/cat<n><pl>]

Translation

Method 1: Create a Translator object and call its translate method.

In [1]: import apertium
In [2]: t = apertium.Translator('eng', 'spa')
In [3]: t.translate('cats')
Out[3]: 'Gatos'

Method 2: Calling translate() directly.

In [1]: import apertium
In [2]: apertium.translate('en', 'spa', 'cats')
Out[2]: 'Gatos'

Installing more modes from other language data

One can also install modes by providing the path to the lang-data:

In [1]: import apertium
In [2]: apertium.append_pair_path('..')