46af45bb3c
* Escape all characters with Markdown significance There are many punctuation characters that sometimes have significance in Markdown; more systematically escape them all (based on a new escape_misc configuration option). A limited attempt is made to limit the escaping of '.' and ')' to the context where they might have Markdown significance (after a number, where they can indicate an ordered list item); no such attempt is made for the other characters (and even that limiting of '.' and ')' may not be entirely safe in all cases, as it's possible the HTML could have the number outside the block being escaped in one go, e.g. `<span>1</span>.`. --------- Co-authored-by: AlexVonB <AlexVonB@users.noreply.github.com>
48 lines
1.4 KiB
Python
48 lines
1.4 KiB
Python
from markdownify import markdownify as md
|
|
|
|
|
|
def test_asterisks():
|
|
assert md('*hey*dude*') == r'\*hey\*dude\*'
|
|
assert md('*hey*dude*', escape_asterisks=False) == r'*hey*dude*'
|
|
|
|
|
|
def test_underscore():
|
|
assert md('_hey_dude_') == r'\_hey\_dude\_'
|
|
assert md('_hey_dude_', escape_underscores=False) == r'_hey_dude_'
|
|
|
|
|
|
def test_xml_entities():
|
|
assert md('&') == r'\&'
|
|
|
|
|
|
def test_named_entities():
|
|
assert md('»') == u'\xbb'
|
|
|
|
|
|
def test_hexadecimal_entities():
|
|
# This looks to be a bug in BeautifulSoup (fixed in bs4) that we have to work around.
|
|
assert md(''') == '\x27'
|
|
|
|
|
|
def test_single_escaping_entities():
|
|
assert md('&amp;') == r'\&'
|
|
|
|
|
|
def text_misc():
|
|
assert md('\\*') == r'\\\*'
|
|
assert md('<foo>') == r'\<foo\>'
|
|
assert md('# foo') == r'\# foo'
|
|
assert md('> foo') == r'\> foo'
|
|
assert md('~~foo~~') == r'\~\~foo\~\~'
|
|
assert md('foo\n===\n') == 'foo\n\\=\\=\\=\n'
|
|
assert md('---\n') == '\\-\\-\\-\n'
|
|
assert md('+ x\n+ y\n') == '\\+ x\n\\+ y\n'
|
|
assert md('`x`') == r'\`x\`'
|
|
assert md('[text](link)') == r'\[text](link)'
|
|
assert md('1. x') == r'1\. x'
|
|
assert md('not a number. x') == r'not a number. x'
|
|
assert md('1) x') == r'1\) x'
|
|
assert md('not a number) x') == r'not a number) x'
|
|
assert md('|not table|') == r'\|not table\|'
|
|
assert md(r'\ <foo> &amp; | ` `', escape_misc=False) == r'\ <foo> & | ` `'
|