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_comparison_equal(self):
me1 = MarkdownElement(
'`',
['text ', MarkdownElement('*', ['text2'], hidden=True)])
me2 = MarkdownElement(
'`',
['text ', MarkdownElement('*', ['text2'], hidden=True)])
self.assertEqual(me1, me2)
def test_comparison_inequal_data_element(self):
me1 = MarkdownElement(
'`',
['text ', MarkdownElement('*', ['text2'], hidden=True)])
me2 = MarkdownElement(
'`',
['text ', MarkdownElement('*', ['text2'], hidden=True)])
me2.data[1].tag = ''
self.assertNotEqual(me1, me2)
def test_format_asserts_no_more_than_one_empty_line(self):
md = MarkdownElement(data=[
'This\n', '\n', 'is\n', '\n', '\n'
'a\n', '\n', '\n', '\n', 'test'
])
expected = MarkdownElement(
data=['This\n', '\n', 'is\n', '\n', 'a\n', '\n', 'test'])
actual = self.mdf.format(md)
self.assertEqual(expected, actual)
def test_comparison_inequal_nested_parent(self):
me1 = MarkdownElement(
'`',
['text ', MarkdownElement('*', ['text2'], hidden=True)])
me2 = MarkdownElement(
'`',
['text ', MarkdownElement('*', ['text2'], hidden=True)])
me2.data[1].parent = None
self.assertNotEqual(me1, me2)
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 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 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_preserve_tree(self):
html = [
'<em>This is</em> ', '<code><em>some code</em> ',
'<code><em>nested inside</em></code> ',
'<em>other code</em></code>. ', '<em>whoa</em>'
]
expected = MarkdownElement(data=[
MarkdownElement('*', ['This is']), ' ',
MarkdownElement('`', [
MarkdownElement('*', ['some code'], hidden=True), ' ',
MarkdownElement(
'`',
[MarkdownElement('*', ['nested inside'], hidden=True)],
hidden=True), ' ',
MarkdownElement('*', ['other code'], hidden=True)
]), '. ',
MarkdownElement('*', ['whoa'])
])
actual = self.md.convert(html, flatten=False)
self.assertEqual(expected, actual)
def test_is_text_only_empty(self):
me = MarkdownElement()
self.assertFalse(me.is_text_only)
def test_default_instance(self):
me = MarkdownElement()
self.assertEqual('', me.tag)
self.assertEqual([], me.data)
self.assertIsNone(None, me.parent)
self.assertFalse(me.hidden)
def test_nondefault_instance(self):
me = MarkdownElement('tag', ['data'], hidden=True)
self.assertEqual('tag', me.tag)
self.assertEqual(['data'], me.data)
self.assertTrue(me.hidden)
def test_parent_add_data(self):
child = MarkdownElement('"', ['text2'])
me = MarkdownElement('`', ['text '])
me.add_data(child)
self.assertEqual(me, child.parent)
def test_tag_supports_list(self):
me = MarkdownElement(['- ', '\n'], ['text'])
expected = '- text\n'
actual = me.flatten()
self.assertEqual(expected, actual)
def test_comparison_different_types(self):
me = MarkdownElement(
'`',
['text ', MarkdownElement('*', ['text2'], hidden=True)])
other = 1
self.assertNotEqual(me, other)
def test_parent_constructor(self):
child = MarkdownElement('"', ['text2'])
me = MarkdownElement('`', ['text ', child])
self.assertEqual(me, child.parent)
def test_is_text_only_has_tag(self):
me = MarkdownElement('tag')
self.assertFalse(me.is_text_only)
def test_is_text_only_has_tag_and_text(self):
me = MarkdownElement('tag', ['stuff'])
self.assertFalse(me.is_text_only)
def test_is_text_only_text_only(self):
me = MarkdownElement(data=['stuff'])
self.assertTrue(me.is_text_only)
def test_flatten_n1(self):
me = MarkdownElement('`', ['text'])
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_respects_hidden_n1(self):
me = MarkdownElement('`', ['text'], hidden=True)
expected = 'text'
actual = me.flatten()
self.assertEqual(expected, actual)
def test_str_calls_flatten(self):
me = MarkdownElement('`', ['text ', MarkdownElement('"', ['text2'])])
expected = '`text "text2"`'
actual = str(me)
self.assertEqual(expected, actual)
