def test_flatten_respects_hidden_n2(self):
me = MarkdownElement(
'`',
['text ', MarkdownElement('*', ['text2'], hidden=True)])
expected = '`text text2`'
actual = me.flatten()
self.assertEqual(expected, actual)
def test_add_data(self):
me = MarkdownElement('`', [])
m2 = MarkdownElement('*', ['text', MarkdownElement('"', ['text2'])])
me.add_data('text3')
me.add_data(m2)
expected = '`text3*text"text2"*`'
actual = me.flatten()
self.assertEqual(expected, actual)
class HtmlToMarkdown(HTMLParser):
def __init__(self):
super().__init__()
self.current_element = None
def convert(self, html, flatten=True):
if isinstance(html, str):
html = [html]
self.current_element = MarkdownElement()
for line in html:
self.feed(line)
if flatten:
return self.current_element.flatten()
else:
return self.current_element
def handle_starttag(self, tag, attrs): # feed() callback
new_element = MarkdownElement()
try:
new_element.tag = [MAP_TAG_START[tag], MAP_TAG_END[tag]]
except KeyError:
new_element.tag = MAP_TAG_START[tag]
self.set_visibility(new_element)
self.current_element.add_data(new_element)
self.current_element = new_element
def set_visibility(self, element):
should_hide = self.current_element.tag in PREFMT_TAGS \
or self.current_element.hidden
element.hidden = should_hide
def handle_endtag(self, tag): # feed() callback
self.current_element = self.current_element.parent
def handle_data(self, data): # feed() callback
data = self.sanitize(data)
self.current_element.data.append(data)
def sanitize(self, data):
return data.replace(u'\xa0', ' ')
def test_tag_supports_list(self):
me = MarkdownElement(['- ', '\n'], ['text'])
expected = '- text\n'
actual = me.flatten()
self.assertEqual(expected, actual)
def test_flatten_respects_hidden_n1(self):
me = MarkdownElement('`', ['text'], hidden=True)
expected = 'text'
actual = me.flatten()
self.assertEqual(expected, actual)
def test_flatten_n2(self):
me = MarkdownElement('`', ['text ', MarkdownElement('"', ['text2'])])
expected = '`text "text2"`'
actual = me.flatten()
self.assertEqual(expected, actual)
def test_flatten_n1(self):
me = MarkdownElement('`', ['text'])
expected = '`text`'
actual = me.flatten()
self.assertEqual(expected, actual)