def test_update_state_check_goal_words():
from trigrams import update_state, initialize_state
old_state = initialize_state()
state = initialize_state()
state['q_goal'] = 'w5'
state['p_goal'] = 'w6'
state['b_goal'] = 'w7'
state['q_word_count'] = 5
state['p_word_count'] = 6
state['b_word_count'] = 7
frmt_word = ' dog'
source_meta = {'proper_names': [],
'quotes': ['s3'],
'parens': ['w8'],
'brackets': []}
also_print, state = update_state(old_state, state, frmt_word, source_meta)
assert also_print == '")]'
assert state['q_word_count'] == 0
assert state['q_sentc_count'] == 0
assert state['q_goal'] == ''
assert not state['in_q']
assert state['p_word_count'] == 0
assert state['p_sentc_count'] == 0
assert state['p_goal'] == ''
assert not state['in_p']
assert state['b_word_count'] == 0
assert state['b_sentc_count'] == 0
assert state['b_goal'] == ''
assert not state['in_b']
def test_update_state_end_sentc_no_q():
from trigrams import update_state, initialize_state
old_state = initialize_state()
state = initialize_state()
old_state['in_s'] = True
state['in_s'] = False
frmt_word = ''
source_meta = {}
also_print, state = update_state(old_state, state, frmt_word, source_meta)
assert also_print == ''
assert state['q_sentc_count'] == 0
def test_check_state_changes():
from trigrams import check_state_changes, initialize_state
old_state = initialize_state()
state = initialize_state()
old_state['in_s'] = True
state['in_p'] = True
state['in_b'] = True
state['in_q'] = True
results = check_state_changes(old_state, state)
assert 'end sentc' in results
assert 'start quote' in results
assert 'start parens' in results
assert 'start brackets' in results
def test_update_state_start_quote():
from trigrams import update_state, initialize_state
old_state = initialize_state()
state = initialize_state()
old_state['in_q'] = False
state['in_q'] = True
frmt_word = ''
source_meta = {'proper_names': [],
'quotes': ['s3'],
'parens': ['w8'],
'brackets': []}
also_print, state = update_state(old_state, state, frmt_word, source_meta)
assert also_print == ''
assert state['q_goal'] == 's3'
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_update_state_space():
from trigrams import update_state, initialize_state
old_state = initialize_state()
state = initialize_state()
old_state['in_q'] = True
old_state['in_p'] = True
old_state['in_b'] = True
state['in_q'] = True
state['in_p'] = True
state['in_b'] = True
frmt_word = ' dog'
source_meta = {'proper_names': [],
'quotes': ['s3'],
'parens': ['w8'],
'brackets': []}
also_print, state = update_state(old_state, state, frmt_word, source_meta)
assert also_print == ''
assert state['q_word_count'] == 1
assert state['p_word_count'] == 1
assert state['b_word_count'] == 1
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'