def test_suffix_before_prefix():
string = "w o r - d h u + s u f"
word, prefix, stem, suffix = split_word_components(string)
assert word == []
assert stem == []
assert prefix == []
assert suffix == []
def test_invalid_word_multiple_affixes():
string = "h e - - l o +"
word, prefix, stem, suffix = split_word_components(string)
assert word == []
assert stem == []
assert prefix == []
assert suffix == []
def test_split_prefix_and_suffix():
string = "p r e + s t e m - s u f"
word, prefix, stem, suffix = split_word_components(string)
assert word == ['p', 'r', 'e', '+', 's', 't', 'e', 'm', '-', 's', 'u', 'f']
assert prefix == [['p', 'r', 'e', '+']]
assert stem == ['s', 't', 'e', 'm']
assert suffix == [['-', 's', 'u', 'f']]
def test_split_suffix():
string = "w o r d - w i n"
word, prefix, stem, suffix = split_word_components(string)
assert word == ['w', 'o', 'r', 'd', '-', 'w', 'i', 'n']
assert prefix == []
assert suffix == [['-', 'w', 'i', 'n']]
assert stem == ['w', 'o', 'r', 'd']
def test_stem_only():
string = "w o r d"
word, prefix, stem, suffix = split_word_components(string)
assert word == ['w', 'o', 'r', 'd']
assert prefix == []
assert suffix == []
assert stem == word
def test_split_prefix():
string = "w o r + d w i"
word, prefix, stem, suffix = split_word_components(string)
assert word == ['w', 'o', 'r', '+', 'd', 'w', 'i']
assert prefix == [['w', 'o', 'r', '+']]
assert suffix == []
assert stem == ['d', 'w', 'i']
def run_test_on_strings(input_list, output_list, object):
for i in range(len(input_list)):
string = input_list[i]
word_as_list, prefix_as_list, stem_as_list, suffix_as_list = \
split_word_components(string)
proceed, preprocessed = \
preprocess(word_as_list, prefix_as_list, stem_as_list, suffix_as_list, object)
if proceed:
assert output_list[i] == postprocess(object.step(preprocessed), object)
else:
assert output_list[i] == "".join(preprocessed)
def test_multiple_prefixes():
string = "w o r + d a n + o t + e a u w p q u a"
word, prefix, stem, suffix = split_word_components(string)
assert word == [
'w', 'o', 'r', '+', 'd', 'a', 'n', '+', 'o', 't', '+', 'e', 'a', 'u',
'w', 'p', 'q', 'u', 'a'
]
assert prefix == [['w', 'o', 'r', '+'], ['d', 'a', 'n', '+'],
['o', 't', '+']]
assert stem == ['e', 'a', 'u', 'w', 'p', 'q', 'u', 'a']
assert suffix == []