wordscape/dictionary.py

import spellchecker
import pathlib
import concurrent
from concurrent.futures import ThreadPoolExecutor
EXECUTOR = ThreadPoolExecutor()
DEFAULT = spellchecker.WordFrequency()
ALTERNATE = spellchecker.WordFrequency()


package_root = pathlib.Path(spellchecker.__file__) / '..'
english = package_root / 'resources' / 'en.json.gz'

extra_words = pathlib.Path(__file__, '..', 'extra_words.txt')


def do_operation(obj, bound_method, *args, **kwargs):
    bound_method(*args, **kwargs)
    return obj


default_load_future = EXECUTOR.submit(
    do_operation, DEFAULT, DEFAULT.load_dictionary, english)
alternate_load_future = EXECUTOR.submit(
    do_operation, ALTERNATE, ALTERNATE.load_text_file, english)


class Dictionary:
    def __init__(self, loader_future: concurrent.futures._base.Future):
        self.future = loader_future
        self._word_frequency = None

    @property
    def word_frequency(self) -> spellchecker.WordFrequency:
        if self._word_frequency is None:
            self._word_frequency = self.future.result()
        return self._word_frequency

    def filter(self, wordlist, excludes=None):
        """Finds words that are in the dictionary"""
        if excludes is None:

            def filter_func(word): return word in self.word_frequency
        else:
            def filter_func(
                word): return word not in excludes and word in self.word_frequency
        return set(filter(filter_func, wordlist))


DEFAULT = Dictionary(default_load_future)
ALTERNATE = Dictionary(alternate_load_future)