def process_expression(expression, max_syll, singletons, verbose, tally=None):
"""Report cases of more than the number of allowed syllable combinations.
Examples are:
1) more than `max_syll` consecutive syllables without spaces or
2) more than `singletons` consecutive single syllables.
Return `tally` in any case:
[0]: more than `max_syll` consecutive syllables without spaces
[1]: more than `singletons` consecutive single syllables
All results are printed to STDOUT."""
if tally == None:
tallly = [0, 0]
divided = expression.split()
consec_sing_count = 0
for item in divided:
try:
length = len(S.separate(item, 'r'))
except TypeError as e:
if verbose:
print(' Skipping {}.'.format(item))
continue
if length > 1:
# Report any case of excessive singletons before resetting counter.
if consec_sing_count > singletons:
if verbose:
print('{}: {} consecutive single syllables.'.
format(expression, consec_sing_count))
tally[1] += 1
consec_sing_count = 0
# Check for excessive number of syllables.
if length > max_syll:
if verbose:
print('{}: {} has syllables numbering {} > {}.'.
format(expression, item, length, max_syll))
tally[0] += 1
else:
consec_sing_count += 1
# Once more, report any case of excessive singletons.
if consec_sing_count > singletons:
if verbose:
print('{}: {} consecutive single syllables.'.
format(expression, consec_sing_count))
tally[1] += 1
return tally
def main(filename='pinyin_syllables_20130728.txt', output=False):
'''Runs never-ending test of separate_pinyin.py.
Use main() for standard syllables; use main('r') for all syllables,
including those with epenthetic -r.
Randomly generates a string of Pīnyīn up to six syllables long,
separates it into component syllables using separate_pinyin.py,
checks whether returned components are the same as the originals or not.
Repeats this process until interrupted with control-c, and then reports the
number of trials done. The process also terminates in the case of a failed
trial.'''
print(
'''Tests of separate_pinyin.py run indefinitely until '''
'''\n\n 1) an error is encountered or'''
'''\n 2) control-c is pressed. '''
'''\n\nIf the latter, statistics are then output.'''
'''\n\nInventory of possible syllables includes''',
end=' ')
if filename == 'r':
print('all those with -r.')
filename = 'pinyin_syllables_with_r_20130728.txt'
else:
print('only those without added -r.')
all_syllables = S.get_py_syllables(filename)
all_vowels, _, _, _, _ = S.get_all_letters()
counter = 0
start_time = time.time()
try:
while True:
counter += 1
string_in_midprocess, original_list = get_random_string(
all_syllables, all_vowels)
finished_list = S.separate(string_in_midprocess, filename)
if finished_list != original_list:
print('Bad!', finished_list, '\n vs.', original_list)
elif output:
print('From {}\n produced {}.'.format(
string_in_midprocess, finished_list))
except KeyboardInterrupt:
total_time = time.time() - start_time
print('\n{:.0f} tests/second'.format(counter / total_time))
print('Total tests run:', counter)
def main(filename='pinyin_syllables_20130728.txt', output=False):
'''Runs never-ending test of separate_pinyin.py.
Use main() for standard syllables; use main('r') for all syllables,
including those with epenthetic -r.
Randomly generates a string of Pīnyīn up to six syllables long,
separates it into component syllables using separate_pinyin.py,
checks whether returned components are the same as the originals or not.
Repeats this process until interrupted with control-c, and then reports the
number of trials done. The process also terminates in the case of a failed
trial.'''
print('''Tests of separate_pinyin.py run indefinitely until '''
'''\n\n 1) an error is encountered or'''
'''\n 2) control-c is pressed. '''
'''\n\nIf the latter, statistics are then output.'''
'''\n\nInventory of possible syllables includes''', end=' ')
if filename == 'r':
print('all those with -r.')
filename = 'pinyin_syllables_with_r_20130728.txt'
else:
print('only those without added -r.')
all_syllables = S.get_py_syllables(filename)
all_vowels, _, _, _, _ = S.get_all_letters()
counter = 0
start_time = time.time()
try:
while True:
counter += 1
string_in_midprocess, original_list = get_random_string(
all_syllables, all_vowels)
finished_list = S.separate(string_in_midprocess, filename)
if finished_list != original_list:
print('Bad!', finished_list, '\n vs.', original_list)
elif output:
print('From {}\n produced {}.'.
format(string_in_midprocess, finished_list))
except KeyboardInterrupt:
total_time = time.time() - start_time
print('\n{:.0f} tests/second'.
format(counter/total_time))
print('Total tests run:', counter)
