def test_format_word_extra_quote():
from trigrams import format_word, initialize_state
state = initialize_state()
state['in_q'] = True
next_word, state = format_word('frog', '"', state)
assert next_word == ''
assert state['in_q']
def test_format_word_out_sentc_end():
from trigrams import format_word, initialize_state
state = initialize_state()
state['in_s'] = False
next_word, state = format_word('frog', '.', state)
assert next_word == '.'
assert not state['in_s']
def test_format_word_in_sentc_alpha():
from trigrams import format_word, initialize_state
state = initialize_state()
state['in_s'] = True
next_word, state = format_word("", 'apple', state)
assert next_word == ' apple'
assert state['in_s']
def test_format_word_em_dash():
from trigrams import format_word, initialize_state
state = initialize_state()
state['in_s'] = True
next_word, state = format_word(u'\u2014', 'really', state)
assert next_word == 'really'
def test_format_word_start_brackets():
from trigrams import format_word, initialize_state
state = initialize_state()
next_word, state = format_word('frog', '[', state)
assert next_word == ' ['
assert state['in_b']
def test_format_word_start_parens():
from trigrams import format_word, initialize_state
state = initialize_state()
next_word, state = format_word('frog', '(', state)
assert next_word == ' ('
assert state['in_p']
def test_format_word_out_sentc_num():
from trigrams import format_word, initialize_state
state = initialize_state()
next_word, state = format_word("", '18', state)
assert next_word == ' 18'
assert not state['in_s']
def test_format_word_nls_yes():
from trigrams import format_word, initialize_state
state = initialize_state()
next_word, state = format_word('hop', '!', state)
assert next_word == '!'
def test_format_word_nls_no():
from trigrams import format_word, initialize_state
state = initialize_state()
state['in_s'] = True
next_word, state = format_word('hop', 'hop', state)
assert next_word == ' hop'