def test_remove_invalid_ipa_characters_invalid_single(self):
values = [
(None, None),
(u"", ([], [])),
(u"f", ([u"f"], [])),
(u"foo", ([u"f", u"o", u"o"], [])),
(u"\u0066\u02BCoo", ([u"\u0066", u"\u02BC", u"o",
u"o"], [])), # single (\u0066 + \u02BC)
(u"f\u031Aoo", ([u"f", u"\u031A", u"o", u"o"], [])),
(u"f\u006e\u0361\u006doo",
([u"f", u"\u006e", u"\u0361", u"\u006d", u"o",
u"o"], [])), # single (\u006e + \u0361 + \u006d)
(u"L", ([], [u"L"])),
(u"LfM", ([u"f"], [u"L", u"M"])),
(u"fLoMo", ([u"f", u"o", u"o"], [u"L", u"M"])),
(u"L\u0066\u02BCMoo", ([u"\u0066", u"\u02BC", u"o",
u"o"], [u"L", u"M"])),
(u"LfM\u02BCoo", ([u"f", u"\u02BC", u"o", u"o"], [u"L", u"M"])),
(u"fL\u031AMoo", ([u"f", u"\u031A", u"o", u"o"], [u"L", u"M"])),
(u"f\u006eL\u0361\u006doo",
([u"f", u"\u006e", u"\u0361", u"\u006d", u"o", u"o"], [u"L"])),
]
for v, e in values:
self.assertEqual(
remove_invalid_ipa_characters(v,
return_invalid=True,
single_char_parsing=True), e)
def command_clean(string, vargs):
"""
Remove characters that are not IPA valid from the given string,
and print the remaining string.
:param str string: the string to act upon
:param dict vargs: the command line arguments
"""
valid_chars, invalid_chars = remove_invalid_ipa_characters(
unicode_string=string,
return_invalid=True,
single_char_parsing=vargs["single_char_parsing"])
print(u"".join(valid_chars))
print_invalid_chars(invalid_chars, vargs)
def command_clean(string, vargs):
"""
Remove characters that are not IPA valid from the given string,
and print the remaining string.
:param str string: the string to act upon
:param dict vargs: the command line arguments
"""
valid_chars, invalid_chars = remove_invalid_ipa_characters(
unicode_string=string,
return_invalid=True,
single_char_parsing=vargs["single_char_parsing"]
)
print(u"".join(valid_chars))
print_invalid_chars(invalid_chars, vargs)
def __init__(self, ipa_chars=None, unicode_string=None, ignore=False, single_char_parsing=False):
self.ipa_chars = []
if ipa_chars is not None:
self.ipa_chars = ipa_chars
elif unicode_string is not None:
if not is_unicode_string(unicode_string):
raise ValueError("The given string is not a Unicode string.")
if (not ignore) and (not is_valid_ipa(unicode_string)):
raise ValueError("The given string contains characters not IPA valid. Use the 'ignore' option to ignore them.")
substrings = remove_invalid_ipa_characters(
unicode_string=unicode_string,
return_invalid=False,
single_char_parsing=single_char_parsing
)
self.ipa_chars = [UNICODE_TO_IPA[substring] for substring in substrings]
def command_check(string, vargs):
"""
Check if the given string is IPA valid.
If the given string is not IPA valid,
print the invalid characters.
:param str string: the string to act upon
:param dict vargs: the command line arguments
"""
is_valid = is_valid_ipa(string)
print(is_valid)
if not is_valid:
valid_chars, invalid_chars = remove_invalid_ipa_characters(
unicode_string=string, return_invalid=True)
print_invalid_chars(invalid_chars, vargs)
def command_check(string, vargs):
"""
Check if the given string is IPA valid.
If the given string is not IPA valid,
print the invalid characters.
:param str string: the string to act upon
:param dict vargs: the command line arguments
"""
is_valid = is_valid_ipa(string)
print(is_valid)
if not is_valid:
valid_chars, invalid_chars = remove_invalid_ipa_characters(
unicode_string=string,
return_invalid=True
)
print_invalid_chars(invalid_chars, vargs)
def test_remove_invalid_ipa_characters_invalid_single(self):
values = [
(None, None),
(u"", ([], [])),
(u"f", ([u"f"], [])),
(u"foo", ([u"f", u"o", u"o"], [])),
(u"\u0066\u02BCoo", ([u"\u0066", u"\u02BC", u"o", u"o"], [])), # single (\u0066 + \u02BC)
(u"f\u031Aoo", ([u"f", u"\u031A", u"o", u"o"], [])),
(u"f\u006e\u0361\u006doo", ([u"f", u"\u006e", u"\u0361", u"\u006d", u"o", u"o"], [])), # single (\u006e + \u0361 + \u006d)
(u"L", ([], [u"L"])),
(u"LfM", ([u"f"], [u"L", u"M"])),
(u"fLoMo", ([u"f", u"o", u"o"], [u"L", u"M"])),
(u"L\u0066\u02BCMoo", ([u"\u0066", u"\u02BC", u"o", u"o"], [u"L", u"M"])),
(u"LfM\u02BCoo", ([u"f", u"\u02BC", u"o", u"o"], [u"L", u"M"])),
(u"fL\u031AMoo", ([u"f", u"\u031A", u"o", u"o"], [u"L", u"M"])),
(u"f\u006eL\u0361\u006doo", ([u"f", u"\u006e", u"\u0361", u"\u006d", u"o", u"o"], [u"L"])),
]
for v, e in values:
self.assertEqual(remove_invalid_ipa_characters(v, return_invalid=True, single_char_parsing=True), e)