def test_has_cjk(self):
result = Chunk('Hello').has_cjk()
self.assertFalse(result,
'should be false when no CJK character is included.')
result = Chunk(u'AとB').has_cjk()
self.assertTrue(result,
'should be true when any CJK character is included.')
def test_wbr_serialize(self):
chunks = ChunkList(Chunk(u'今日は'), Chunk(u'ご飯を'), Chunk(u'食べます。'))
result = chunks.wbr_serialize()
expected = ('<span style="word-break: keep-all;">'
u'今日は<wbr></wbr>ご飯を<wbr></wbr>食べます。'
'</span>')
self.assertEqual(result, expected,
'Chunks should be separated by WBR tags.')
def test_swap(self):
old_chunks = self.chunks[0:2]
new_chunk = Chunk('ijk')
self.chunks.swap(old_chunks, new_chunk)
expected = ['ijk', 'fgh']
self.assertEqual(expected, [chunk.word for chunk in self.chunks],
'Old chunks should be replaced with the new chunk.')
def test_is_open_punct(self):
puncts = [u'。', u'、', u'「', u'」', u'（', u'）', u'[', u']', u'(', u')']
expected = [
False, False, True, False, True, False, True, False, True, False
]
results = [Chunk(c).is_open_punct() for c in puncts]
self.assertListEqual(expected, results,
'Open punctuation marks should be detected.')
def setUp(self):
self.chunks = ChunkList(Chunk('ab', dependency=None),
Chunk('cde', dependency=True),
Chunk('fgh', dependency=False))
def test_span_serialize(self):
chunks = ChunkList(Chunk('Hello'), Chunk.space(), Chunk(u'今天'),
Chunk(u'天气'), Chunk(u'很好'))
attributes = {'class': 'foo'}
expected = ('<span>'
'Hello '
u'<span class="foo">今天</span>'
u'<span class="foo">天气</span>'
u'<span class="foo">很好</span>'
'</span>')
result = chunks.span_serialize(attributes, None)
self.assertEqual(result, expected,
'The chunks should be compiled to a HTML code.')
chunks = ChunkList(Chunk('Hey<'), Chunk('<script>alert(1)</script>'),
Chunk('>guys'))
attributes = {'class': 'foo'}
expected = ('<span>'
'Hey<<script>alert(1)</script>>guys'
'</span>')
result = chunks.span_serialize(attributes, None)
self.assertEqual(result, expected,
'HTML tags included in a chunk should be encoded.')
chunks = ChunkList(Chunk(u'去年'), Chunk(u'インフルエンザに'), Chunk(u'かかった。'))
attributes = {'class': 'foo'}
expected = ('<span>'
u'<span class="foo">去年</span>'
u'インフルエンザに'
u'<span class="foo">かかった。</span>'
'</span>')
result = chunks.span_serialize(attributes, 6)
self.assertEqual(
result, expected,
'Chunks that exceed the max length should not be enclosed by a span.'
)
def test_insert_breaklines(self):
chunks = ChunkList(Chunk(u'これが '), Chunk('Android'))
chunks._insert_breaklines()
self.assertEqual(
[u'これが', '\n', 'Android'], [chunk.word for chunk in chunks],
'Trailing spaces in CJK chunk should be converted to breaklines.')