AutoCorrect

A Python module that provides a dictionary and a simple learning algorithm to autocorrect words & texts.

---

Getting started

To get started with this module you've got to create an AutoCorrect dictionary.

import AutoCorrect
my_dictionary = AutoCorrect.Dictionary()

The first thing to do with a dictionary is learn it some words.

# It can learn a single word.
my_dictionary.learn_word('foobar')

# Or it can learn from a piece of text.
my_dictionary.learn_text('Lorem ipsum dolor')

Once that is done, you can start using the dictionary to autocorrect words.

# Oops, I typed 'foobar' incorrect.
mistake = 'fooabr'

# Lets see if my_dictionary can correct my mistake!
correction = my_dictionary.correct_word(mistake)

---

General API

.Dictionary()

Create a new instance of the dictionary provided by this Python module.

import AutoCorrect
my_dictionary = AutoCorrect.Dictionary()

Parameters

Parameter	Required	Description
alphabet	no	A list of characters that can exist in the dictionary

.Import()

Import an AutoCorrect dictionary from a file (or string) and get a new dictionary instance with the contents of the imported dictionary.

import AutoCorrect
my_dictionary = AutoCorrect.Import(path='./foobar.json')

Parameters

Parameter	Required	Description
file	any	A file object of the file that should be imported.
path	any	A string of the path to the file that should be imported.
string	any	A string of a serialized AutoCorrect Dictionary.

Dictionary API

.learn_text()

Learn all the words from a string of text to the dictionary. This method expects only one of the parameters to be provided.

my_dictionary.learn_text(text='Hello world!')

Parameter	Required	Description
text	any	A string of the text to correct.
file	any	The file with the text to correct.
path	any	A string of the path to the file with the text to correct.

.learn_word()

Learn a single word to the dictionary.

my_dictionary.learn_word('foobar')

Parameters

Parameter	Required	Description
word	yes	The word that should be learned by the dictionary.
follows	no	A word that tends to follow the given word.
leads	no	A word that tends to lead the given word.

.correct_text()

Automatically correct a complete text based on the contents of the dictionary. This will scan through the text and replace unknown words with the best alternative it can find. This method expects only one of the parameters to be provided.

corrected_text = my_dictionary.correct_text(text='Helol worl!')
print(corrected_text) # Gives 'Hello world!'

Parameter	Required	Description
text	any	A string of the text to correct.
file	any	The file with the text to correct.
path	any	A string of the path to the file with the text to correct.

.correct_word()

Automatically correct a single word based on the contents of the dictionary.

corrected_word = my_dictionary.correct_word('fobar')
print(corrected_word) # Gives 'foobar'

Parameter	Required	Description
word	yes	A string of the word to correct.
follows	no	A string of the word that is in front of word.
leads	no	A string of the word that comes after word.

.find_longer_words()

Find all words in the dictionary that have a certain prefix. If no prefix is provided this returns the complete dictionary.

suggestions = my_dictionary.find_longer_words('foo')
print(suggestions) # Gives ['foobar']

Parameter	Required	Description
prefix	no	A string with the prefix of choice.

.find_similar_words()

Find all similar words for a given string known to the dictionary using a combination of multiple different algorithms.

This method will return a list suggestions formatted in tuples of the following structure: (word, rating), ordered descending based on the rating.

suggestions = my_dictionary.find_similar_words('fobar')
print(suggestions) # Suggests 'foobar'

Parameter	Required	Description
word	yes	A string of the word to correct.
follows	no	A string of the word that is in front of word.
leads	no	A string of the word that comes after word.

.find_word()

Find a word in the dictionary and get a tuple with some information about it.

word = my_dictionary.find_word('foobar')
print(word) # Gives a tuple with information about 'foobar'

Parameter	Required	Description
word	yes	A string of the word to correct.

.get_dictionary()

Get all the words known to the dictionary.

dictionary = my_dictionary.get_dictionary()
print(dictionary) # Gives a list of all the words in the dictionary

.suggestion_feedback()

Give manual feedback on the suggestions offered by the dictionary. This will increase the rating of the suggested word the next time the dictionary is asked to correct the incorrect word.

my_dictionary.suggestion_feedback('nol', 'not')

Parameters

Parameter	Required	Description
incorrect	yes	The word for which the suggestion was offered.
suggestion	yes	The suggested word that can be found in the dictionary.

.unlearn

Unlearn something the Dictionary knows. It is possible to unlearn a word by only providing a string of that word as the first argument. If the word wasn't found, a LookupError will be thrown.

If you want to unlearn something specific about a word, you should provide it as a string for the thing you want to unlearn (for the details, see below).

# To unlearn a word
my_dictionary.unlearn('foobar')

# To unlearn something about a word
my_dictionary.unlearn('foo', leads='bar')

Parameters

Parameter	Required	Description
word	yes	The word to unlearn (something about).
feedback	no	A piece feedback on the word to forget.
follows	no	A word the dictionary expects in front of word to forget.
leads	no	A word the dictionary expects after word to forget.

.export()

Export the dictionary to a file so it can be imported at a later point in time.

my_dictionary.export(path='./foobar.json')

Parameter	Required	Description
file	any	The file to write the dictionary to.
path	any	A string of the path to the file to write the dictionary to.

---

Testing

In order to run the tests for this module use the command:

$ python -m unittest discover

---

Copyright © 2016-2017 Eric Cornelissen | MIT license