def testSplitWithPaddedOutput(self, texts, expected, ragged_rank=None):
input_tensor = ragged.constant_value(_nested_encode(texts, "UTF-8"),
ragged_rank=ragged_rank,
dtype=bytes)
result = ragged.unicode_split(input_tensor,
"UTF-8").to_tensor(default_value="")
self.assertAllEqual(np.array(expected, dtype=bytes), result)
def testSplitWithSparseOutput(self, texts, expected):
input_tensor = np.array(_nested_encode(texts, "UTF-8"), dtype=bytes)
result = ragged.unicode_split(input_tensor, "UTF-8").to_sparse()
self.assertIsInstance(result, sparse_tensor.SparseTensor)
self.assertAllEqual(expected.indices, result.indices)
self.assertAllEqual(expected.values, result.values)
self.assertAllEqual(expected.dense_shape, result.dense_shape)
def testBasicSplit(self, texts, ragged_rank=None):
input_tensor = ragged.constant_value(_nested_encode(texts, "UTF-8"),
ragged_rank=ragged_rank,
dtype=bytes)
result = ragged.unicode_split(input_tensor, "UTF-8")
expected = _nested_splitchars(texts, "UTF-8")
self.assertRaggedEqual(expected, result)
def testDocstringExamples(self):
texts = [s.encode("utf8") for s in [u"G\xf6\xf6dnight", u"\U0001f60a"]]
codepoints1 = ragged.unicode_split(texts, "UTF-8")
codepoints2, offsets = ragged.unicode_split_with_offsets(
texts, "UTF-8")
self.assertRaggedEqual(codepoints1, [[
b"G", b"\xc3\xb6", b"\xc3\xb6", b"d", b"n", b"i", b"g", b"h", b"t"
], [b"\xf0\x9f\x98\x8a"]])
self.assertRaggedEqual(codepoints2, [[
b"G", b"\xc3\xb6", b"\xc3\xb6", b"d", b"n", b"i", b"g", b"h", b"t"
], [b"\xf0\x9f\x98\x8a"]])
self.assertRaggedEqual(offsets, [[0, 1, 3, 5, 6, 7, 8, 9, 10], [0]])
def testExceptions(self, exception=None, message=None, **args):
with self.assertRaisesRegexp(exception, message):
self.evaluate(ragged.unicode_split(**args))
def testSplitWithDifferentEncodings(self, encoding, texts):
expected = _nested_splitchars(texts, encoding)
input_tensor = constant_op.constant(_nested_encode(texts, encoding))
result = ragged.unicode_split(input_tensor, encoding)
self.assertRaggedEqual(expected, result)
def testErrorModes(self, expected=None, **args):
result = ragged.unicode_split(**args)
self.assertRaggedEqual(expected, result)
def testVectorSplit(self):
text = constant_op.constant([u"仅今年前".encode("UTF-8"), b"hello"])
chars = ragged.unicode_split(text, "UTF-8")
expected_chars = [[c.encode("UTF-8") for c in u"仅今年前"],
[c.encode("UTF-8") for c in u"hello"]]
self.assertRaggedEqual(chars, expected_chars)
def testScalarSplit(self):
text = constant_op.constant(u"仅今年前".encode("UTF-8"))
chars = ragged.unicode_split(text, "UTF-8")
self.assertAllEqual(chars, [c.encode("UTF-8") for c in u"仅今年前"])
