def fix_spells():
""" Runs through the document, finds all the bad words, then
tries to find fixed versions of them.
"""
lang = get_lang()
checker = spell_checker.AspellSpellChecker(lang)
db = document_builder.SpellcheckDocMaker(checker)
db.make_word_fix_doc('text/clean')
def simple_clean():
""" Simple cleanup for testing algorithm. """
os.system('cp text/raw/* text/simple_clean/')
lang = get_lang()
checker = spell_checker.AspellSpellChecker(lang)
# checker.fixer = spell_checker.SimpleEnglishSpellFixer()
db = document_builder.SpellcheckDocMaker(checker)
db.remove_possible_headers('text/simple_clean')
lm = line_manager.LineManager(
spell_checker.FileConfiguredSpellChecker(lang))
lm.load('text/simple_clean')
lm.quick_fix()
lm.join_lines()
# lm.write_pages('text/simple_clean', False)
# db.make_word_fix_doc('text/simple_clean')
lm.write_pages('text/simple_clean', True)
def test_checkables(self):
to_test = (
('bad', 'company', [
'badcompany',
'bacompany',
]),
('bad-', 'company', [
'badcompany',
]),
("bad'", 'company', [
'badcompany',
]),
('bad', 'Company', []),
(u'ba\u00E0', 'company', [
u'ba\u00E0company',
'bacompany',
]),
('bad', u'\u00E0ompany', [
u'bad\u00E0ompany',
u'ba\u00E0ompany',
]),
(
'ba',
'company',
[],
),
(
'bad',
'com',
[
'badcom',
'bacom',
],
),
(
'bad',
'co',
[],
),
)
db = document_builder.SpellcheckDocMaker(
spell_checker.StubSpellChecker([]))
for word1, word2, expected in to_test:
self.assertEqual(spell_checker.joinables(word1, word2), expected)
def test_fixed_words(self):
sc = spell_checker.StubSpellChecker([
'Cantrip',
'government',
'bomb',
'born',
'bod',
"he'll",
'What',
'hiss',
'different',
])
db = document_builder.SpellcheckDocMaker(sc)
for test, expected in test_expected(
'{}/test_spellcheck/fix_spelling'.format(PATH)):
self.assertEqual(
db.fixed_words((test, )).values()[0], [
expected,
])
def remove_headers():
lang = get_lang()
dict_ = './dict.{}.pws'.format(lang)
checker = spell_checker.AspellSpellChecker(lang, dict_)
db = document_builder.SpellcheckDocMaker(checker)
db.remove_possible_headers('text/clean')
def cross_line_fixes():
lang = get_lang()
checker = spell_checker.AspellSpellChecker(lang)
db = document_builder.SpellcheckDocMaker(checker)
db.make_line_join_doc('text/clean')
def proper_names():
lang = get_lang()
dict_ = './dict.{}.pws'.format(lang)
checker = spell_checker.AspellSpellChecker(lang, dict_)
db = document_builder.SpellcheckDocMaker(checker)
db.make_possible_proper_name_doc('text/clean')
def possible_headers():
lang = get_lang()
checker = spell_checker.AspellSpellChecker(lang)
db = document_builder.SpellcheckDocMaker(checker)
db.possible_headers('text/raw')
def page_info():
lang = get_lang()
checker = spell_checker.AspellSpellChecker(lang)
db = document_builder.SpellcheckDocMaker(checker)
db.page_image_info('text/raw', 'images/pages')