def test_split_using_dictionary_key_one_single(self):
d = dict()
d[u"a"] = 1
d[u"b"] = 2
d[u"c"] = 4
values = [
(None, None),
(u"", []),
(u"aza", [u"a", u"z", u"a"]),
(u"aaba", [u"a", u"a", u"b", u"a"]),
(u"acaba", [u"a", u"c", u"a", u"b", u"a"]),
]
for v, e in values:
self.assertEqual(split_using_dictionary(v, d, 1, single_char_parsing=True), e)
def test_split_using_dictionary_key_one_single(self):
d = dict()
d[u"a"] = 1
d[u"b"] = 2
d[u"c"] = 4
values = [
(None, None),
(u"", []),
(u"aza", [u"a", u"z", u"a"]),
(u"aaba", [u"a", u"a", u"b", u"a"]),
(u"acaba", [u"a", u"c", u"a", u"b", u"a"]),
]
for v, e in values:
self.assertEqual(
split_using_dictionary(v, d, 1, single_char_parsing=True), e)
def test_split_using_dictionary(self):
d = dict()
d[u"a"] = 1
d[u"ba"] = 2
d[u"b"] = 3
d[u"c"] = 4
d[u"ca"] = 5
values = [
(None, None),
(u"", []),
(u"aza", [u"a", u"z", u"a"]),
(u"aaba", [u"a", u"a", u"ba"]),
(u"acaba", [u"a", u"ca", u"ba"]),
]
for v, e in values:
self.assertEqual(split_using_dictionary(v, d, 2, single_char_parsing=False), e)
def can_map_ipa_string(self, ipa_string):
"""
Return ``True`` if the mapper can map all the IPA characters
in the given IPA string.
:param IPAString ipa_string: the IPAString to be parsed
:rtype: bool
"""
canonical = [(c.canonical_representation, ) for c in ipa_string]
split = split_using_dictionary(canonical,
self,
self.max_key_length,
single_char_parsing=False)
for sub in split:
if not sub in self.ipa_canonical_representation_to_mapped_str:
return False
return True
def test_split_using_dictionary(self):
d = dict()
d[u"a"] = 1
d[u"ba"] = 2
d[u"b"] = 3
d[u"c"] = 4
d[u"ca"] = 5
values = [
(None, None),
(u"", []),
(u"aza", [u"a", u"z", u"a"]),
(u"aaba", [u"a", u"a", u"ba"]),
(u"acaba", [u"a", u"ca", u"ba"]),
]
for v, e in values:
self.assertEqual(
split_using_dictionary(v, d, 2, single_char_parsing=False), e)
def map_ipa_string(self,
ipa_string,
ignore=False,
return_as_list=False,
return_can_map=False):
"""
Convert the given IPAString to a string
containing the corresponding ASCII IPA representation.
:param IPAString ipa_string: the IPAString to be parsed
:param bool ignore: if ``True``, ignore Unicode characters that are not IPA valid
:param bool return_as_list: if ``True``, return as a list of strings, one for each IPAChar,
instead of their concatenation (single str)
:param bool return_can_map: if ``True``, return a pair ``(bool, str)``, where the first element
says if the mapper can map all the IPA characters in the given IPA string,
and the second element is either ``None`` or the mapped string/list
:rtype: str or (bool, str) or (bool, list)
"""
acc = []
can_map = True
canonical = [(c.canonical_representation, ) for c in ipa_string]
split = split_using_dictionary(canonical,
self,
self.max_key_length,
single_char_parsing=False)
for sub in split:
try:
acc.append(
self.ipa_canonical_representation_to_mapped_str[sub])
except KeyError:
if ignore:
can_map = False
else:
raise ValueError(
"The IPA string contains an IPA character that is not mapped: %s"
% sub)
mapped = acc if return_as_list else u"".join(acc)
if return_can_map:
return (can_map, mapped)
return mapped