plural inflections

[DimitriPapadopoulos]

Module inflect generates the plural inflection of English words.

This could be used to detect whether the plural inflection of a suggestion is available in the dictionary. If it is not, chances are the plural inflection of the associated misspelling is missing too.

This is convoluted, but using inflect directly on misspellings might not work well - to be tested - and will mean more words to look into.

[DimitriPapadopoulos]

See rationale here:
#2077 (comment)

[DimitriPapadopoulos]

Here is a proof of concept:

#!/usr/bin/python3

import os
import inflect

DICTIONARY_DIR = "/my/path/codespell/codespell_lib/data"

backwards = {}

for dictionary in os.listdir(DICTIONARY_DIR):
    root, ext = os.path.splitext(dictionary)
    if root.startswith("dictionary") and ext == ".txt":
        dictionary = os.path.join(DICTIONARY_DIR, dictionary)
        with open(dictionary) as f:
            for line in f:
                key, data = line.split("->")

                key = key.lower()
                key = key.strip()

                data = data.lower()
                for word in data.split(','):
                    word = word.strip()
                    if word:
                        backwards.setdefault(word, []).append(key)


p = inflect.engine()

for word in backwards:
    plural = p.plural(word)
    if plural not in backwards:
        print(word, backwards[word])

Most words are not nouns. The difficult part is to single out nouns. Any way, approx. 1 out of 10 reported words are nouns. For example, the following are missing from the dictionaries:
infarctions->infractions
aminators->animators, laminators,
reorganisations->reorganizations
webaservers->webservers, web servers, (I cannot find any occurrence in Google, except in a "valid" webaservers.net domain)
webages->webpages
weaponaries->weaponries

[DimitriPapadopoulos]

To find nouns, see Library to grammatically classify English words (nouns, verbs, adverbs, etc). Use the Python module nltk.